An event integration method, device, equipment and computer readable storage medium

Abstract

The application provides an event integration method, device and equipment and a computer readable storage medium, which are applied to various scenes such as cloud technology, artificial intelligence, intelligent transportation and vehicle-mounted devices. The method comprises the following steps: obtaining an event to be integrated, obtaining at least one topic to be integrated, each topic to be integrated comprising at least one topic event; based on selection logic, selecting one or more of semantic similarity, string graph similarity and question and answer similarity to obtain a target similarity between the event to be integrated and each topic to be integrated, the string graph similarity being the similarity of a graph feature corresponding to a key string, and the question and answer similarity being the similarity of a question and answer feature; based on the target similarity, determining a target topic to which the event to be integrated belongs from the at least one topic to be integrated; and integrating the event to be integrated into the target topic to obtain an event context comprising the event to be integrated and the at least one topic event. Through the application, the accuracy of event integration can be improved.

Description

Technical Field

[0001] This application relates to information processing technology in the field of computer applications, and more particularly to an event integration method, apparatus, device, and computer-readable storage medium. Background Technology

[0002] For topics that have been ongoing for a long time (often consisting of multiple events that have already occurred), when the latest developments are obtained, they need to be integrated into the corresponding topic to form an event timeline that includes the latest developments. This allows users to intuitively understand the development of the event through the event timeline.

[0003] Generally, to integrate the latest developments into a topic, clustering is typically used. This involves incrementally clustering the latest developments with the topic, and then determining the topic to which the latest development belongs based on the cluster centers and thresholds. However, when integrating events using incremental clustering, the accuracy of clustering is relatively low, resulting in a low accuracy in determining the assigned topic. Consequently, the accuracy of integrating the latest developments into a topic is also low. Summary of the Invention

[0004] This application provides an event integration method, apparatus, device, and computer-readable storage medium that can improve event integration efficiency.

[0005] The technical solution of this application embodiment is implemented as follows:

[0006] This application provides an event integration method, including:

[0007] Obtain events to be integrated and at least one topic to be integrated, wherein each topic to be integrated includes at least one topic event;

[0008] Based on selection logic, one or more of semantic similarity, string graph similarity, and question-answer similarity are selected to obtain the target similarity between the event to be integrated and each topic to be integrated. Here, semantic similarity refers to the similarity in terms of semantic features, string graph similarity refers to the similarity in terms of graph features corresponding to key strings, and question-answer similarity refers to the similarity in terms of question-answer features.

[0009] Based on the target similarity, determine the target topic to which the event to be integrated belongs from at least one of the topics to be integrated;

[0010] The events to be integrated are integrated into the target topic to obtain an event context that includes the events to be integrated and at least one event of the topic.

[0011] This application provides an event integration device, including:

[0012] The information acquisition module is used to acquire events to be integrated and to acquire at least one topic to be integrated, wherein each topic to be integrated includes at least one topic event;

[0013] The similarity acquisition module is used to select one or more from semantic similarity, string graph similarity and question-answer similarity based on selection logic to obtain the target similarity between the event to be integrated and each topic to be integrated. The semantic similarity refers to the similarity in terms of semantic features, the string graph similarity refers to the similarity in terms of graph features corresponding to key strings, and the question-answer similarity refers to the similarity in terms of question-answer features.

[0014] A topic determination module is used to determine the target topic to which the event to be integrated belongs from at least one of the topics to be integrated, based on the target similarity.

[0015] The event integration module is used to integrate the events to be integrated into the target topic to obtain an event outline including the events to be integrated and at least one event of the topic.

[0016] In this embodiment of the application, the selection logic includes one or more of the following: selection order, acquisition speed, accuracy, topic size, number of selections, topic type, model training scale, model applicability scope, and model applicability scale. The selection order is determined based on the priority of similarity. The acquisition speed is the speed at which similarity is acquired. The accuracy is the degree of accuracy of the similarity. The topic type is the content format of the topic to be integrated. The topic size is the size of at least one topic to be integrated. The model training scale is the training data size corresponding to the network model used to acquire each type of similarity.

[0017] In this embodiment of the application, when the selection logic includes the selection order, the similarity acquisition module is further configured to, based on the selection order, sequentially select a first predetermined number of similarities from the descending order of priority of semantic similarity, string graph similarity, and question-and-answer similarity; obtain a comparison result between the first predetermined number of similarities and a similarity threshold; when the comparison result is a similarity result between the event to be integrated and the topic to be integrated, determine the first predetermined number of similarities as the target similarity between the event to be integrated and the topic to be integrated; when the comparison result is an undetermined similarity result between the event to be integrated and the topic to be integrated, continue to select the remaining similarities based on the selection order until the similarity result is determined or three of the semantic similarity, string graph similarity, and question-and-answer similarity are selected, and determine the selected multiple similarities as the target similarity between the event to be integrated and the topic to be integrated, wherein the remaining similarities are the similarities among the semantic similarity, string graph similarity, and question-and-answer similarity other than the first predetermined number of similarities.

[0018] In this embodiment of the application, when the selection logic includes the acquisition speed and the topic size, the similarity acquisition module is further configured to, when the topic size is greater than a set size, sequentially select a second set number of similarities from the descending order of the acquisition speed of the semantic similarity, the string graph similarity, and the question-answer similarity to obtain the target similarity between the event to be integrated and the topic to be integrated; when the topic size is less than or equal to the set size, sequentially select a third set number of similarities from the descending order of the acquisition speed of the semantic similarity, the string graph similarity, and the question-answer similarity to obtain the target similarity between the event to be integrated and the topic to be integrated, wherein the second set number is less than the third set number.

[0019] In this embodiment of the application, when the target similarity includes multiple types of semantic similarity, string graph similarity, and question-answering similarity, the topic determination module is further configured to determine the weight ratio of various similarities in the target similarity based on accuracy; based on the weight ratio, fuse the multiple similarities in the target similarity to obtain a discriminative similarity; and select the topic to be integrated corresponding to the highest discriminative similarity from at least one topic to be integrated to obtain the target topic to which the event to be integrated belongs.

[0020] In this embodiment, the semantic similarity includes semantic self-attention similarity. The similarity acquisition module is further configured to acquire the semantic features to be integrated corresponding to the event to be integrated, and the semantic features of each topic event in the topic to be integrated; enhance the semantic features to be integrated based on the difference identifiers of the event to be integrated and the topic events to obtain a first enhanced semantic feature, and enhance the semantic features of the topic events based on the difference identifiers to obtain a second enhanced semantic feature; form a semantic feature sequence by combining the first enhanced semantic feature with at least one second enhanced semantic feature corresponding to the topic to be integrated, and determine the semantic self-attention similarity based on the self-attention information between two sequence units in the semantic feature sequence.

[0021] In this embodiment of the application, the similarity acquisition module is further configured to: acquire a first sub-semantic similarity between the title of each topic event and the title of the event to be integrated in each topic to be integrated; and determine an average first sub-semantic similarity and a maximum first sub-semantic similarity based on the first sub-semantic similarity; acquire a second sub-semantic similarity between the topic event key string corresponding to each topic event and the event key string corresponding to the event to be integrated in each topic to be integrated; and determine an average second sub-semantic similarity and a maximum second sub-semantic similarity based on the second sub-semantic similarity; acquire a third sub-semantic similarity between the topic key string corresponding to each topic to be integrated and the event key string to be integrated; and determine the average first sub-semantic similarity, the maximum first sub-semantic similarity, the average second sub-semantic similarity, the maximum second sub-semantic similarity, and the third sub-semantic similarity as the semantic statistical similarity between the event to be integrated and each topic to be integrated.

[0022] In this embodiment, the first sub-semantic similarity, the second sub-semantic similarity, and the third sub-semantic similarity are obtained through a semantic statistical similarity model. The event integration device further includes a model training module for acquiring training samples, wherein the training samples include a first string sample, a second string sample, and labeled similarity. A first semantic branch in the semantic statistical similarity model to be trained is used to acquire a first estimated semantic corresponding to the first string sample; a second semantic branch in the semantic statistical similarity model to be trained is used to acquire a second estimated semantic corresponding to the second string sample; and based on the comparison result between the first estimated semantic and the second estimated semantic, the estimated similarity between the first string sample and the second string sample is determined. Based on the difference between the estimated similarity and the labeled similarity, backpropagation is performed in the semantic statistical similarity model to be trained to obtain the semantic statistical similarity model.

[0023] In this embodiment of the application, the similarity acquisition module is further configured to: determine each sub-topic event key string corresponding to at least one topic event as a graph node in each topic to be integrated; construct an edge between two graph nodes corresponding to two sub-topic event key strings belonging to the same topic event to obtain a first key string graph; construct a second key string graph based on the key string of the event to be integrated corresponding to the event to be integrated; and determine the string graph similarity between the event to be integrated and each topic to be integrated based on the comparison result between the vector representation of the first key string graph and the vector representation of the second key string graph.

[0024] In this embodiment of the application, the similarity acquisition module is further configured to combine a sequence of statements to be answered based on the title, topic key strings, and the event to be integrated for each of the topics to be integrated; obtain answer information for the sequence of statements to be answered; and determine the question-and-answer similarity between the event to be integrated and each of the topics to be integrated based on the answer information.

[0025] In this embodiment of the application, the information acquisition module is further configured to acquire, in the topic library, the matching result of the topic key string corresponding to each topic and the event to be integrated, wherein the topic library includes multiple topics; based on the matching result, when it is determined that at least one sub-topic key string in the topic key string matches the event to be integrated, the topic corresponding to the matching result is determined as the topic to be integrated that matches the event to be integrated; and at least one topic to be integrated that matches the event to be integrated is acquired from the topic library.

[0026] In this embodiment of the application, the information acquisition module is further configured to acquire a topic event key string corresponding to each topic event in at least one topic event corresponding to each topic in the topic library; count the number of topic events corresponding to each sub-topic event key string in the topic event key string; and combine the sub-topic event key strings with a fourth set number of maximum topic event counts into a topic key string corresponding to each topic.

[0027] In this embodiment of the application, the information acquisition module is further configured to perform entity recognition on each topic event to obtain entity key strings corresponding to preset entity types; perform string weight analysis on each topic event to obtain action key strings; and determine the topic event key strings based on one or both of the entity key strings and the action key strings.

[0028] In this embodiment of the application, the information acquisition module is further configured to acquire the number of entity key strings corresponding to the entity key string; when the number of entity key strings is less than a fifth preset number, the entity key string and the action key string are combined into the topic event key string; when the entity key string is greater than or equal to the fifth preset number, the entity key string is determined as the topic event key string.

[0029] In this embodiment, the event integration device further includes an event display module for presenting a search control; in response to a first search operation applied to the search control, a simplified event network corresponding to the event network and a presentation control corresponding to the simplified event network are presented, wherein the simplified event network belongs to the event network, and the presentation control is used to present the event network; in response to a presentation operation applied to the presentation control, the event network is presented, wherein each event in the presented event network includes an event title and an event time, and the event is any one of the event to be integrated and at least one of the topic events; in response to a view operation applied to the event title or the event time, event details information are presented.

[0030] In this embodiment of the application, the event display module is further configured to present the last information to be presented of the target event, wherein the target event is any one of the events to be integrated and at least one of the topic events included in the event network; in the recommendation area corresponding to the last information to be presented, the remaining events in the event network associated with the target event are presented, wherein the remaining events are any events in the event network other than the target event; and in response to a second search operation on the remaining events, the detailed information of the remaining events is presented.

[0031] This application provides an event integration device, including:

[0032] Memory, used to store executable instructions;

[0033] The processor, when executing executable instructions stored in the memory, implements the event integration method provided in the embodiments of this application.

[0034] This application provides a computer-readable storage medium storing executable instructions for implementing the event integration method provided in this application when executed by a processor.

[0035] This application provides a computer program product, including a computer program or instructions, which, when executed by a processor, implement the event integration method provided in this application.

[0036] The embodiments of this application have at least the following beneficial effects: When determining the target topic to which the event to be integrated belongs in at least one topic to be integrated, the target topic is determined by judging the target similarity between the event to be integrated and each topic to be integrated. That is, the target topic is determined by directly comparing the target similarity between the event to be integrated and each topic to be integrated. Since the target similarity includes one or more of semantic similarity, string graph similarity and question-answering similarity, the obtained target similarity can accurately determine whether each topic to be integrated is the target topic to which the event to be integrated belongs. Therefore, when the event to be integrated is integrated into the target topic, the accuracy of event integration can be improved. Attached Figure Description

[0037] Figure 1 This is an optional architecture diagram of the event integration system provided in the embodiments of this application;

[0038] Figure 2 This is provided by the embodiments of this application. Figure 1 A schematic diagram illustrating an exemplary component structure of a server in a given context;

[0039] Figure 3 This is an optional flowchart illustrating the event integration method provided in the embodiments of this application;

[0040] Figure 4 This is another optional flowchart illustrating the event integration method provided in the embodiments of this application;

[0041] Figure 5 This is another optional flowchart illustrating the event integration method provided in the embodiments of this application;

[0042] Figure 6 This is a schematic diagram illustrating an exemplary event timeline provided in an embodiment of this application;

[0043] Figure 7 This is another exemplary event timeline presentation diagram provided in the embodiments of this application;

[0044] Figure 8 This is another optional flowchart illustrating the event integration method provided in the embodiments of this application;

[0045] Figure 9 This is an exemplary schematic diagram of news topic recall provided in an embodiment of this application;

[0046] Figure 10 This is an exemplary diagram illustrating how to determine whether a news topic is related to the latest developments, provided in an embodiment of this application.

[0047] Figure 11aThis is a schematic diagram of an exemplary model for obtaining vector semantic similarity provided in an embodiment of this application;

[0048] Figure 11b This is a schematic diagram of another exemplary model for obtaining vector semantic similarity provided in the embodiments of this application;

[0049] Figure 12 This is a schematic diagram of an exemplary keyword diagram provided in an embodiment of this application;

[0050] Figure 13 This is a schematic diagram of another exemplary keyword diagram provided in an embodiment of this application;

[0051] Figure 14 This is an exemplary schematic diagram illustrating the acquisition of semantic similarity in question-and-answer processing, provided by an embodiment of this application.

[0052] Figure 15 This is a schematic diagram illustrating an exemplary feature importance provided in an embodiment of this application. Detailed Implementation

[0053] To make the objectives, technical solutions, and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings. The described embodiments should not be regarded as limitations on this application. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0054] In the following description, references are made to “some embodiments,” which describe a subset of all possible embodiments. However, it is understood that “some embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict.

[0055] In the following description, the terms “first, second, third, fourth, fifth” are used merely to distinguish similar objects and do not represent a specific ordering of objects. It is understood that “first, second, third, fourth, fifth” may be interchanged in a specific order or sequence where permitted, so that the embodiments of this application described herein can be implemented in an order other than that illustrated or described herein.

[0056] Unless otherwise defined, all technical and scientific terms used in the embodiments of this application have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used in the embodiments of this application is for the purpose of describing the embodiments of this application only and is not intended to limit this application.

[0057] Before providing a further detailed description of the embodiments of this application, the nouns and terms involved in the embodiments of this application will be explained, and the nouns and terms involved in the embodiments of this application shall be interpreted as follows.

[0058] 1) Artificial Intelligence (AI) is the theory, methods, technology, and application systems that use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results. In other words, AI is a comprehensive technology within computer science used to capture the essence of intelligence and produce new intelligent machines that can react in a way similar to human intelligence. Furthermore, AI is used to study the design principles and implementation methods of various intelligent machines, enabling them to possess perception, reasoning, and decision-making capabilities. Moreover, AI technology is a comprehensive discipline involving a wide range of fields, encompassing both hardware and software technologies. Fundamental AI technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing technology, operating / interactive systems, and mechatronics. AI software technologies mainly include computer vision, speech processing, natural language processing, and machine learning (ML) / deep learning.

[0059] 2) Natural Language Processing (NLP) is an important field within computer science and artificial intelligence. It refers to the study of theories and methods that enable effective communication between humans and computers using natural language. Therefore, NLP is a science integrating linguistics, computer science, and mathematics; consequently, research in NLP involves natural language, that is, the language people use in daily life, thus NLP is closely related to linguistics. NLP technologies typically include Machine Reading Comprehension (MRC), text processing, semantic understanding, machine translation, question answering, and knowledge graphs.

[0060] 3) Machine learning is a multidisciplinary field involving probability theory, statistics, approximation theory, convex analysis, and algorithm complexity theory, among others. It studies how computers can simulate or implement human learning behavior to acquire new knowledge or skills and reorganize existing knowledge structures to continuously improve their performance. Machine learning is the core of artificial intelligence and the fundamental way to endow computers with intelligence. Its applications span all areas of artificial intelligence. Machine learning and deep learning typically include techniques such as artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning, and instructional learning.

[0061] 4) Machine reading comprehension is a natural language processing task; it is used to answer a question by machine after reading an article, given an article and a question based on the article; wherein, the article and the question are the sequence of statements to be answered in the embodiments of this application.

[0062] 5) Graph Convolutional Network (GCN) is used to compute graph representations. The corresponding data processed is graph-structured data. A graph is a data format used to represent key string networks, social networks, communication networks, and protein molecular networks, etc. Nodes in a graph represent individuals in the network, and edges represent the connections between individuals. In this embodiment, the vector representations of the first key string graph and the second key string graph can be obtained through a graph convolutional network.

[0063] 6) Named Entity Recognition (NER), also known as entity recognition, entity segmentation, and entity extraction, is used to locate and classify named entities in text into predefined categories, such as people, organizations, locations, time expressions, quantities, currency values, percentages, etc. Typically, the task of NER is to identify three main categories (entity, time, and number) and seven subcategories (person names, organization names, place names, time, date, currency, and percentage) of named entities in the text to be processed. In this embodiment, named entity recognition is used to obtain entities of preset entity types, such as person names and place names.

[0064] Generally, to integrate the latest events into a topic, clustering is typically used. This involves incrementally clustering the latest events with the topics, and then determining the topic to which the latest event belongs based on the cluster centers and thresholds. However, this incremental clustering method suffers from computational overhead that increases with the number of topics, resulting in low efficiency in event integration.

[0065] Based on this, embodiments of this application provide an event integration method, apparatus, device, and computer-readable storage medium, which can improve event integration efficiency and reduce the computational resource consumption of event integration. The following describes exemplary applications of the event integration device provided in this application. The event integration device provided in this application can be implemented as various types of terminals such as smartphones, smartwatches, laptops, tablets, desktop computers, smart home appliances, set-top boxes, smart in-vehicle devices, portable music players, personal digital assistants, dedicated messaging devices, smart voice interaction devices, portable gaming devices, and smart speakers, or it can be implemented as a server. The following will describe exemplary applications when the device is implemented as a server.

[0066] See Figure 1 , Figure 1 This is an optional architecture diagram of the event integration system provided in the embodiments of this application; as shown... Figure 1 As shown, to support an event integration application, in the event integration system 100, terminal 200 (terminals 200-1 and 200-2 are shown as examples) connects to server 400 (event integration device) via network 300. Network 300 can be a wide area network (WAN), a local area network (LAN), or a combination of both. Additionally, the event integration system 100 includes a database 500 for providing data support (at least one topic to be integrated) to server 400; and... Figure 1 The example shown illustrates a scenario where the database 500 is independent of the server 400. However, the database 500 can also be integrated into the server 400, and this embodiment does not limit this to any particular case.

[0067] Terminal 200 is used to obtain the event context from server 400 via network 300 and display the event context on a graphical interface.

[0068] Server 400 is used to acquire events to be integrated and at least one topic to be integrated, wherein each topic to be integrated includes at least one topic event. Based on selection logic, it selects one or more of semantic similarity, string-graph similarity, and question-answer similarity to obtain the target similarity between the events to be integrated and each topic to be integrated. Semantic similarity refers to the similarity in terms of semantic features, string-graph similarity refers to the similarity in terms of graph features corresponding to key strings, and question-answer similarity refers to the similarity in terms of question-answer features. Based on the target similarity, it determines the target topic to which the events to be integrated belong from the at least one topic to be integrated. It integrates the events to be integrated into the target topic to obtain an event context including the events to be integrated and at least one topic event. It is also used to send the event context to terminal 200 via network 300.

[0069] In some embodiments, server 400 may be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms. Terminal 200 may be a smartphone, tablet, laptop, desktop computer, smart speaker, smartwatch, smart voice interaction device, smart home appliance, or in-vehicle terminal, but is not limited to these. Terminals and servers can be directly or indirectly connected via wired or wireless communication, which is not limited in this embodiment.

[0070] See Figure 2 , Figure 2 This is provided by the embodiments of this application. Figure 1 A schematic diagram illustrating an exemplary component structure of a server in a given context; Figure 2 The server 400 shown includes at least one processor 410, a memory 450, and at least one network interface 420; in some embodiments of this application, the server 400 also includes a user interface 430. The various components of the server 400 are coupled together via a bus system 440. It is understood that the bus system 440 is used to implement communication between these components. In addition to a data bus, the bus system 440 also includes a power bus, a control bus, and a status signal bus. However, for clarity, in… Figure 2 The general labeled all buses as Bus System 440.

[0071] Processor 410 can be an integrated circuit chip with signal processing capabilities, such as a general-purpose processor, a digital signal processor (DSP), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among them, the general-purpose processor can be a microprocessor or any conventional processor, etc.

[0072] User interface 430 includes one or more output devices 431 that enable the presentation of media content, including one or more speakers and / or one or more visual displays. User interface 430 also includes one or more input devices 432, including user interface components that facilitate user input, such as a keyboard, mouse, microphone, touch screen display, camera, other input buttons and controls.

[0073] The memory 450 may be removable, non-removable, or a combination thereof. Exemplary hardware devices include solid-state storage, hard disk drives, optical disk drives, etc. The memory 450 may optionally include one or more storage devices physically located away from the processor 410.

[0074] The memory 450 may include volatile memory or non-volatile memory, or both. The non-volatile memory may be read-only memory (ROM), and the volatile memory may be random access memory (RAM). The memory 450 described in this application embodiment is intended to include any suitable type of memory.

[0075] In some embodiments of this application, memory 450 is capable of storing data to support various operations. Examples of such data include programs, modules, and data structures, or subsets or supersets thereof, as illustrated below.

[0076] Operating system 451 includes system programs for handling various basic system services and performing hardware-related tasks, such as the framework layer, core library layer, driver layer, etc., for implementing various basic business functions and handling hardware-based tasks;

[0077] The network communication module 452 is used to reach other computer devices via one or more (wired or wireless) network interfaces 420, exemplary network interfaces 420 including: Bluetooth, Wi-Fi, and Universal Serial Bus (USB), etc.

[0078] Presentation module 453 is configured to enable the presentation of information (e.g., a user interface for operating peripheral devices and displaying content and information) via one or more output devices 431 associated with user interface 430 (e.g., a display screen, a speaker, etc.).

[0079] The input processing module 454 is used to detect and translate one or more user inputs or interactions from one or more input devices 432.

[0080] In some embodiments of this application, the event integration device provided in this application can be implemented in software. Figure 2An event integration device 455 stored in memory 450 is shown. This device can be software in the form of programs and plug-ins, and includes the following software modules: information acquisition module 4551, similarity acquisition module 4552, topic determination module 4553, event integration module 4554, model training module 4555, and event display module 4556. These modules are logically connected and can therefore be arbitrarily combined or further separated according to their implemented functions. The functions of each module will be described below.

[0081] In some other embodiments of this application, the event integration device provided in the embodiments of this application can be implemented in hardware. As an example, the event integration device provided in the embodiments of this application can be a processor in the form of a hardware decoding processor, which is programmed to execute the event integration method provided in the embodiments of this application. For example, the processor in the form of a hardware decoding processor can be one or more application-specific integrated circuits (ASICs), DSPs, programmable logic devices (PLDs), complex programmable logic devices (CPLDs), field-programmable gate arrays (FPGAs), or other electronic components.

[0082] The event integration method provided in this application will be described below with reference to exemplary applications and implementations of the event integration device provided in the embodiments of this application. Furthermore, the event integration method provided in the embodiments of this application can be applied to various scenarios such as cloud technology, artificial intelligence, smart transportation, and vehicle-mounted systems.

[0083] See Figure 3 , Figure 3 This is an optional flowchart illustrating the event integration method provided in the embodiments of this application, which will be combined with... Figure 3 The steps shown are explained.

[0084] S301. Obtain the events to be integrated and at least one topic to be integrated.

[0085] In this embodiment, the event integration device acquires the events to be integrated, thus obtaining the events to be integrated. This can be achieved by the event integration device detecting events, or by the event integration device receiving events sent by other devices, etc., and this embodiment does not limit the scope of the acquisition. Furthermore, the event integration device acquires the topics to be integrated, thus obtaining at least one topic to be integrated.

[0086] It should be noted that the "event to be integrated" refers to an event that needs to be integrated, and an event is used to describe information about what happened, such as a news event or a highlight event. Furthermore, the event to be integrated can be a recently developed event or a historical event. Historical events refer to events that occurred after the corresponding event time, and this application embodiment does not limit this. In addition, the event to be integrated includes at least text information, and may also include at least one of audio / video, images, and tables. Furthermore, at least one topic to be integrated can be all topics in the database, or topics that may be related to the event to be integrated and filtered from the database, etc., and this application embodiment does not limit this. Moreover, a topic to be integrated is an event theme, a collection of related events, including at least one topic event, and a topic event is also an event.

[0087] S302. Based on the selection logic, select one or more from semantic similarity, string graph similarity and question-answer similarity to obtain the target similarity between the event to be integrated and each topic to be integrated.

[0088] In this embodiment of the application, the event integration device determines whether each topic to be integrated is the topic to which the event to be integrated belongs by comparing the target similarity between the event to be integrated and each topic to be integrated.

[0089] It should be noted that target similarity refers to the probability that the event to be integrated belongs to each topic to be integrated. Furthermore, target similarity can be determined from one or more aspects, including semantic similarity, string-graph similarity, and question-and-answer similarity, and is determined based on selection logic. Semantic similarity refers to similarity in terms of semantic features, string-graph similarity refers to similarity in terms of graph features corresponding to key strings, and question-and-answer similarity refers to similarity in terms of question-and-answer features. Selection logic is the basis for the event integration device to select from semantic similarity, string-graph similarity, and question-and-answer similarity. The event integration device, based on the selection logic, selects one or more of semantic similarity, string-graph similarity, and question-and-answer similarity to obtain the target similarity, including: the event integration device, based on the selection logic, selects one of semantic similarity, string-graph similarity, and question-and-answer similarity to obtain the target similarity; or, the event integration device, based on the selection logic, selects at least two of semantic similarity, string-graph similarity, and question-and-answer similarity to obtain the target similarity.

[0090] It should also be noted that the selection logic includes one or more of the following: selection order, acquisition speed, accuracy, topic size, number of selections, topic type, model training scale, and model applicable scale. The selection order is determined based on the priority of similarity, which can be based on one or both of accuracy and time consumption; the acquisition speed is the speed at which similarity is acquired, and the acquisition speed can be determined based on one or both of feature extraction time and feature extraction method (parallel or serial); the accuracy is the degree of accuracy of the similarity, which can be determined based on one or both of the characteristics of the features used in the similarity acquisition process or the accuracy of the corresponding network model; the topic type is the content form of the topic to be integrated. For example, when the content form is in image form, string graph similarity and question-answer similarity can be selected as target similarities; when the content form is in text form, one or more of semantic similarity, string graph similarity, and question-answer similarity including semantic similarity can be selected as target similarities; the topic size is the size of at least one topic to be integrated, which can be determined based on one or more of the number of topics to be integrated and the content volume of the topics to be integrated; the model training size is the training data size corresponding to the network model used to acquire each type of similarity; the model applicable size is the maximum amount of data that the network model used to acquire each type of similarity can carry; the model applicable scope is the data form corresponding to the network model used to acquire each type of similarity.

[0091] In this embodiment, the event integration device compares the features of the graph structure corresponding to the event to be integrated with the features of the graph structure corresponding to each topic to be integrated to obtain string-graph similarity. The event integration device can construct questions and articles for machine reading comprehension based on the events to be integrated and each topic to be integrated, and determine the question-and-answer similarity corresponding to the answer information through the interaction between the questions and articles. Furthermore, semantic similarity, string-graph similarity, and question-and-answer similarity are similarities obtained from different dimensions.

[0092] S303. Based on target similarity, determine the target topic to which the event to be integrated belongs from at least one topic to be integrated.

[0093] In this embodiment, when at least one topic to be integrated is a single topic, the event integration device can determine whether this single topic is a target topic by judging the target similarity and a similarity threshold. Furthermore, when at least one topic to be integrated is a topic possibly associated with the event to be integrated, selected from a database, the event integration device can directly identify this single topic as the target topic. When at least one topic to be integrated consists of multiple topics, the event integration device can identify the topic that best matches the event to be integrated from among the at least one topic, and identify the topic that best matches the event to be integrated as the topic to which the event to be integrated belongs, i.e., the target topic. It can also identify the topic corresponding to the maximum target similarity greater than the similarity threshold as the target topic.

[0094] It should be noted that the event integration device obtains at least one target similarity between the event to be integrated and at least one topic to be integrated by acquiring the target similarity between the event to be integrated and each topic to be integrated; then, it determines the target topic from at least one topic to be integrated based on at least one target similarity, or determines the target topic from at least one topic to be integrated based on the comparison result of at least one target similarity and a similarity threshold. This application embodiment does not limit this.

[0095] S304. Integrate the events to be integrated into the target topic to obtain an event timeline that includes the events to be integrated and at least one topic event.

[0096] In this embodiment of the application, the event integration device treats the event to be integrated as a topic event within a target topic, integrating it into at least one topic event included in the target topic, thereby obtaining an event context including the event to be integrated and at least one topic event. Here, the event context refers to the process of events described for the target topic.

[0097] Understandably, when determining the target topic to which an event belongs within at least one topic to be integrated, it is done by judging the target similarity between the event and each topic. That is, the target topic is determined directly by comparing the target similarity between the event and each topic. Since target similarity includes one or more of semantic similarity, string graph similarity, and question-answering similarity, the obtained target similarity can accurately determine whether each topic belongs to the target topic of the event. Therefore, when integrating the event to be integrated into the target topic, the accuracy of event integration is improved. Furthermore, when the event integration device uses multiple methods of semantic similarity, string graph similarity, and question-answering similarity to determine the target topic from at least one topic to be integrated, it is a process based on multi-dimensional heterogeneous features. Therefore, the accuracy and effectiveness of the obtained target topics are high, thus improving the accuracy of event integration.

[0098] See Figure 4 , Figure 4 This is another optional flowchart illustrating the event integration method provided in the embodiments of this application; as shown below. Figure 4 As shown in the embodiments of this application, when the selection logic includes a selection order, S302 can be implemented through S3021 to S3024; that is, the event integration device selects one or more from semantic similarity, string graph similarity and question-answer similarity based on the selection logic to obtain the target similarity between the event to be integrated and each topic to be integrated, including S3021 to S3024. The steps are described below.

[0099] S3021. Based on the selection order, select a first set number of similarities in descending order of priority from semantic similarity, string graph similarity and question-answer similarity.

[0100] It should be noted that the first set number of similarities includes one or more of semantic similarity, string graph similarity, and question-answering similarity.

[0101] For example, the event integration device can first select the question-answer similarity and semantic similarity with the highest accuracy. If the result can be determined, the selection continues. If the result cannot be determined, the selection ends. If the result cannot be determined, the character graph similarity is selected. Alternatively, the event integration device can first select the question-answer similarity with the shortest time consumption. If the result can be determined, the selection ends. If the result cannot be determined, the similarity is selected from semantic similarity and string graph similarity.

[0102] S3022. Obtain the comparison results of the first set number of similarities and the similarity threshold.

[0103] It should be noted that the similarity threshold may include a first set number of character similarity thresholds, and the first set number of character similarity thresholds corresponds one-to-one with the first set number of similarities.

[0104] S3023. When the comparison result is a similarity result between the event to be integrated and the topic to be integrated, the first set number of similarities is determined as the target similarity between the event to be integrated and the topic to be integrated.

[0105] It should be noted that the similarity result between the event to be integrated and the topic to be integrated refers to whether the event to be integrated and the topic to be integrated are similar or dissimilar.

[0106] S3024. When the comparison result is an undetermined similarity result between the event to be integrated and the topic to be integrated, continue to select the remaining similarities based on the selection order until a similarity result is determined or three of the semantic similarity, string graph similarity and question-answer similarity are selected. Then, the selected multiple similarities are determined as the target similarity between the event to be integrated and the topic to be integrated.

[0107] It should be noted that "pending similarity results" refers to situations where it is impossible to determine whether the event to be integrated is similar to or dissimilar to the topic to be integrated; "remaining similarity" refers to the similarity scores among semantic similarity, string-graph similarity, and question-answering similarity, excluding the first set number of similarities. Furthermore, "multiple selected similarities" refers to all similarities selected across all possible selection attempts.

[0108] In this embodiment of the application, when the selection logic includes acquisition speed and topic size, S302 can also be implemented through S3025 and S3026; that is, the event integration device selects one or more from semantic similarity, string graph similarity and question-answer similarity based on the selection logic to obtain the target similarity between the event to be integrated and each topic to be integrated, including S3025 and S3026. The steps are described below.

[0109] S3025. When the topic size is larger than the set size, select the second set number of similarities in descending order of the acquisition speed of semantic similarity, string graph similarity and question-answer similarity to obtain the target similarity between the event to be integrated and the topic to be integrated.

[0110] It should be noted that if the event integration device determines that the scale of the topic is larger than the set scale, it indicates that at least one topic to be integrated is large in scale, and a small number of similarity with faster acquisition speeds (selecting the second set number of similarity in descending order of acquisition speed) are needed to determine the result.

[0111] S3026. When the topic size is less than or equal to the set size, select the third set number of similarities in descending order of the acquisition speed of semantic similarity, string graph similarity and question-answer similarity to obtain the target similarity between the event to be integrated and the topic to be integrated.

[0112] It should be noted that if the event integration device determines that the topic size is less than or equal to the set size, it indicates that at least one topic to be integrated is small in size, and more similarity with faster acquisition speed (selecting the third set number of similarity in descending order of acquisition speed) is needed to determine the result; in addition, the second set number is less than the third set number.

[0113] In this embodiment of the application, when the target similarity includes multiple types such as semantic similarity, string graph similarity and question-answer similarity, S303 can be implemented by S3031 to S3033; that is, the event integration device determines the target topic to which the event to be integrated belongs from at least one topic to be integrated based on the target similarity, including S3031 to S3033. Each step is described below.

[0114] S3031. Determine the weight ratio of various similarities in target similarity based on accuracy.

[0115] It should be noted that the event integration device determines a weight that is positively correlated with the accuracy for each of the selected similarities, thus obtaining the weight ratio among the various similarities in the target similarity; whereby the weight ratio represents the relationship between the weights corresponding to each similarity.

[0116] S3032. Based on the weighting ratio, multiple similarities in the target similarity are fused to obtain the discriminative similarity.

[0117] It should be noted that the event integration device, based on weight allocation, merges each similarity in the target similarity with its corresponding weight. After the fusion of all similarities in the target similarity is completed, the final similarity used to determine whether the event to be integrated and the topic to be integrated are similar is obtained; the final similarity used to determine whether the event to be integrated and the topic to be integrated are similar is called the discrimination similarity.

[0118] S3033. From at least one topic to be integrated, select the topic to be integrated corresponding to the highest discrimination similarity to obtain the target topic to which the event to be integrated belongs.

[0119] It should be noted that the event integration device can directly determine the topic to be integrated corresponding to the highest discrimination similarity as the target topic to which the event to be integrated belongs; it can also compare the highest discrimination similarity with a threshold and then determine whether to directly determine the topic to be integrated corresponding to the highest discrimination similarity as the target topic to which the event to be integrated belongs; etc., the embodiments of this application do not specifically limit this.

[0120] In this embodiment, semantic similarity includes one or both of semantic self-attention similarity and semantic statistical similarity. Semantic self-attention similarity is determined based on the self-attention between the event to be integrated and the topic event. Semantic statistical similarity is obtained by comparing the vector semantic features of the title, keywords, and body text of the event to be integrated with the vector semantic features of the title, keywords, and related information of each topic event to be integrated. Here, semantic self-attention similarity is obtained through the following steps: the event integration device acquires the semantic features to be integrated corresponding to the event to be integrated, and the semantic features of each topic event in the topic to be integrated; based on the distinguishing identifiers of the event to be integrated and the topic events, the semantic features to be integrated are enhanced to obtain a first enhanced semantic feature, and the semantic features of the topic events are enhanced based on the distinguishing identifiers to obtain a second enhanced semantic feature; the first enhanced semantic feature and at least one second enhanced semantic feature corresponding to the topic to be integrated are combined to form a semantic feature sequence, and the semantic self-attention similarity is determined based on the self-attention information between two sequence units in the semantic feature sequence.

[0121] It should be noted that the self-attention information between two sequence units refers to the self-attention between the event to be integrated and any topic event.

[0122] It should also be noted that when semantic similarity includes both semantic self-attention similarity and semantic statistical similarity, one possible implementation method corresponding to S3021 to S3024 includes: the event integration device selects semantic self-attention similarity and question-answer similarity (a first predetermined number of similarities). If, after comparing semantic self-attention similarity and question-answer similarity with their respective sub-similarity thresholds, it is determined that the event to be integrated is dissimilar or similar to the topic event, the process ends. However, if, after comparing semantic self-attention similarity and question-answer similarity with their respective sub-similarity thresholds, it is impossible to determine whether the event to be integrated is dissimilar or similar to the topic event, then semantic statistical similarity and string graph similarity are selected for further discrimination. Here, based on accuracy and acquisition speed, the priority can be determined in descending order as question-answer similarity, semantic self-attention similarity, string graph similarity, and semantic statistical similarity; and, question-answer similarity, semantic self-attention similarity, string graph similarity, and semantic statistical similarity represent a transition from precision to breadth in that order.

[0123] In the embodiments of this application, semantic statistical similarity can be obtained through steps S30211 to S30214. Each step is described below.

[0124] S30211. In each topic to be integrated, obtain the first sub-semantic similarity between the title of each topic event and the title of the event to be integrated, and determine the average first sub-semantic similarity and the maximum first sub-semantic similarity based on the first sub-semantic similarity.

[0125] In this embodiment of the application, the event integration device determines semantic statistical similarity from one or more of the following: the similarity between the title of the event to be integrated and the title of each topic event in each topic to be integrated; the similarity between the key string of the event to be integrated and the key string of each topic event in each topic to be integrated; the similarity between the key string of the event to be integrated and the key string of each topic to be integrated; and the similarity between the event to be integrated and each topic to be integrated.

[0126] Here, for each topic event in each topic to be integrated, the event integration device obtains the similarity between the title of the topic event and the title of the event to be integrated, thus obtaining the first sub-semantic similarity. Therefore, for each topic to be integrated, at least one first sub-semantic similarity can be obtained. The event integration device calculates the average value of the at least one first sub-semantic similarity to obtain the average first sub-semantic similarity. The event integration device selects the largest first sub-semantic similarity from the at least one first sub-semantic similarity to obtain the maximum first sub-semantic similarity.

[0127] S30212. In each topic to be integrated, obtain the second sub-semantic similarity between the topic event key string corresponding to each topic event and the key string of the event to be integrated corresponding to the event to be integrated, and determine the average second sub-semantic similarity and the maximum sub-second semantic similarity based on the second sub-semantic similarity.

[0128] In this embodiment of the application, the event integration device obtains the similarity between the key string of the topic event and the key string of the event to be integrated for each topic event in each topic to be integrated, thus obtaining the second sub-semantic similarity. Therefore, for each topic to be integrated, at least one second sub-semantic similarity can be obtained. The event integration device calculates the average value of the at least one second sub-semantic similarity to obtain the average second sub-semantic similarity. The event integration device selects the largest second sub-semantic similarity from at least two first sub-semantic similarities to obtain the maximum second sub-semantic similarity.

[0129] It should be noted that the keyword string for the topic event is the keyword string for the topic event; the keyword string for the event to be integrated is the keyword string for the event to be integrated.

[0130] S30213. Obtain the third sub-semantic similarity between the topic key string and the event key string corresponding to each topic to be integrated.

[0131] In this embodiment of the application, the event integration device obtains the similarity between the topic key string and the event key string to be integrated, thus obtaining the third sub-semantic similarity.

[0132] S30214. The average first sub-semantic similarity, the maximum first sub-semantic similarity, the average second sub-semantic similarity, the maximum second sub-semantic similarity, and the third sub-semantic similarity are determined as the semantic statistical similarity between the event to be integrated and each topic to be integrated.

[0133] It should be noted that the event integration device can determine at least one of the following as the semantic statistical similarity between the event to be integrated and each topic to be integrated: average first sub-semantic similarity, maximum first sub-semantic similarity, average second sub-semantic similarity, maximum second sub-semantic similarity, and third sub-semantic similarity.

[0134] In this embodiment of the application, the first sub-semantic similarity in S30211, the second sub-semantic similarity in S30212, and the third sub-semantic similarity in S30213 can all be obtained by a semantic statistical similarity model. The semantic statistical similarity model is used to obtain the degree of similarity between text pairs in terms of semantic features. The semantic statistical similarity model is trained by S305 to S307. Each step is described below.

[0135] S305. Obtain training samples, which include a first string sample, a second string sample, and labeled similarity.

[0136] It should be noted that the training samples refer to the data samples used to train the semantic statistical similarity model. The first string sample and the second string sample are text pairs whose similarity in terms of semantic features is to be determined. The labeled similarity is the actual similarity between the first string sample and the second string sample in terms of semantic features.

[0137] S306. Using the first semantic branch in the semantic statistical similarity model to be trained, obtain the first estimated semantic corresponding to the first string sample. Using the second semantic branch in the semantic statistical similarity model to be trained, obtain the second estimated semantic corresponding to the second string sample. Based on the comparison results between the first estimated semantic and the second estimated semantic, determine the estimated similarity between the first string sample and the second string sample.

[0138] In this embodiment, the event integration device initializes the parameters of the model structure, thus obtaining a semantic statistical similarity model to be trained. This model includes a first semantic branch and a second semantic branch. Next, the event integration device uses the first semantic branch to obtain the semantics corresponding to the first string sample, thus obtaining the first estimated semantics. Then, the event integration device uses the second semantic branch to obtain the semantics corresponding to the second string sample, thus obtaining the second estimated semantics. Finally, the similarity model in the semantic statistical similarity model to be trained is used to determine the degree of similarity between the first string sample and the second string sample, thus obtaining the estimated similarity. Here, the similarity model compares the first and second estimated semantics and determines the estimated similarity between the first and second string samples based on the comparison result.

[0139] It should be noted that the semantic statistical similarity model to be trained is a model to be trained to obtain the similarity of semantic features of text pairs; and the semantic statistical similarity model to be trained adopts a dual-tower structure (first semantic branch and second semantic branch), with each semantic branch in the dual-tower structure used to obtain semantic features.

[0140] Understandably, the semantic statistical similarity model to be trained can improve the efficiency of obtaining estimated similarity by using a dual-tower structure to acquire semantic features.

[0141] S307. Based on the difference between the estimated similarity and the labeled similarity, backpropagation is performed in the semantic statistical similarity model to be trained to obtain the semantic statistical similarity model.

[0142] In this embodiment, the event integration device adjusts the parameters in the semantic statistical similarity model to be trained based on the difference between the estimated similarity and the labeled similarity, in order to train the semantic statistical similarity model. Here, the event integration device adjusts the parameters by performing backpropagation in the semantic statistical similarity model to be trained. The training process of the semantic statistical similarity model to be trained is an iterative training process, and the semantic statistical similarity model to be trained after training is the semantic statistical similarity model.

[0143] In the embodiments of this application, the string graph similarity can be obtained through S30221 to S30223. Each step is described below.

[0144] S30221. In each topic to be integrated, determine the key string of each sub-topic event corresponding to at least one topic event as a graph node, and build an edge between the two graph nodes corresponding to the key strings of two sub-topic events belonging to the same topic event to obtain the first key string graph.

[0145] It should be noted that in at least one topic event within each topic to be integrated, the topic event key string corresponding to each topic event includes one or more sub-topic event key strings. Here, the event integration device treats a sub-topic event key string as a graph node and iterates through all the obtained graph nodes. If any two graph nodes are found to be related to the same topic event, an edge is built between the two graph nodes. If the two graph nodes are found to be related to the same topic event, no edge is built between the two graph nodes. Finally, at the end of the traversal, the obtained graph structure is the first key string graph.

[0146] S30222. Construct a second key string graph based on the key strings of the events to be integrated.

[0147] It should be noted that the event integration device constructs a graph structure corresponding to the events to be integrated based on the construction method of the first key string graph: the event integration device treats each sub-key string of the key string of the event to be integrated as a graph node, and builds edges between any two graph nodes, thus obtaining the second key string graph.

[0148] S30223. Based on the comparison results between the vector representations of the first key string graph and the vector representations of the second key string graph, determine the string graph similarity between the event to be integrated and each topic to be integrated.

[0149] In this embodiment of the application, the event integration device obtains the vector representation of the first key string graph and the vector representation of the second key string graph; then, it compares the vector representation of the first key string graph and the vector representation of the second key string graph, and determines the string graph similarity between the event to be integrated and each topic to be integrated based on the comparison result between the vector representation of the first key string graph and the vector representation of the second key string graph.

[0150] In the embodiments of this application, the question-answer similarity can be obtained through S30231 to S30233. Each step is described below.

[0151] S30231. Based on the title, key strings of the topic, and events to be integrated for each topic to be integrated, combine the sequence of statements to be answered.

[0152] It should be noted that, in order to determine whether an event to be integrated belongs to a topic to be integrated through question-and-answer interaction, the event integration device constructs question-and-answer statements corresponding to each topic to be integrated and the event to be integrated, thus obtaining a sequence of statements to be answered. Specifically, the event integration device combines the title of each topic to be integrated, the topic keyword string, and the event to be integrated according to the preset sentence structure of the question-and-answer statements. The resulting combination is the constructed question-and-answer statement; for example, is the question "Is the event to be integrated a development of the title of the topic to be integrated, with the keyword string 'topic keyword string'?"; another example is: Is the next sentence a development of the title of the topic to be integrated, with the keyword string 'topic keyword string', and the event to be integrated?

[0153] S30232. Obtain the answer information for the sequence of statements to be answered.

[0154] In this embodiment, the event integration device determines the corresponding question and article in the sequence of statements to be answered based on the questions and articles in machine reading comprehension. It then performs low-level processing on the determined articles and questions to convert the text into digital codes. Next, it determines the semantic relationship between the articles and questions based on the digital codes, and obtains the features of the determined questions by combining the semantic analysis results of the articles. It also obtains the features of the determined articles by combining the semantic analysis results of the questions. Finally, the event integration result is based on the representation information of the determined questions, the features of the determined articles, and the type of the answer to obtain the output answer information.

[0155] It should be noted that the answer information refers to whether the event to be integrated belongs to each topic to be integrated. It can be "yes" (the event to be integrated belongs to the topic to be integrated), "no" (the event to be integrated does not belong to the topic to be integrated), or the possibility that the event to be integrated belongs to the topic to be integrated, etc. This application embodiment does not limit this.

[0156] S30233. Based on the answer information, determine the question-answer similarity between the event to be integrated and each topic to be integrated.

[0157] It should be noted that the event integration device determines the probability that the event to be integrated belongs to each topic to be integrated based on the answer information, and defines the determined probability as the question-answer similarity between the event to be integrated and each topic to be integrated.

[0158] In this embodiment of the application, the event integration device obtains at least one topic to be integrated in S301, including S3011 to S3013. Each step is described below.

[0159] S3011. In the topic library, obtain the matching results of the topic key string corresponding to each topic and the event to be integrated. The topic library includes multiple topics.

[0160] In this embodiment, the event integration device can obtain a pre-set topic library. After obtaining the event to be integrated, the event integration device determines the topic to which the event belongs from the topic library, and then integrates the event into the topic. Here, the event integration device first matches each topic in the topic library with the event to be integrated, matching the topic keyword string corresponding to each topic with the event to be integrated.

[0161] It should be noted that the topic keyword string is the keyword string of the topic. The topic library includes multiple topics, each topic being the subject of an event; and each topic in the topic library includes at least one topic event, the topic events included in different topics may be the same or different; and at least one topic event refers to an event that occurs at different time periods and is associated with the topic, thus there is a chronological order between at least one topic event.

[0162] S3012. Based on the matching results, when at least one sub-topic key string in the topic key string matches the event to be integrated, the topic corresponding to the matching result is determined as the topic to be integrated that matches the event to be integrated.

[0163] S3013. Obtain at least one topic from the topic library that matches the event to be integrated.

[0164] It should be noted that the event integration device determines the topic matching result between each topic's key string and the event to be integrated. If the key string matches at least one sub-topic key string, then the topic corresponding to that matching result is determined to be a topic to be integrated that matches the event. Conversely, if the key string does not match the event, then the topic is determined not to be a topic to be integrated. Here, when the event integration device obtains multiple matching results between multiple topics and the event to be integrated, it can retrieve at least one topic to be integrated that matches the event from the topic library. If it determines that no topic to be integrated matches the event, a new topic including the event to be integrated will be constructed, and the new topic will be updated in the topic library.

[0165] It is understandable that, in the process of integrating events to be integrated into their respective target topics, the process first involves matching the events to be integrated with the key strings of the topic to recall at least one topic that may be related to the events to be integrated. Then, based on the similarity between the events to be integrated and each topic to be integrated, the target topic to which the events to be integrated belong is accurately determined from at least one topic to be integrated. Therefore, by adopting the recall-similarity classification method, this embodiment of the application can quickly integrate events to be integrated into target topics, and the computation time of the integration process is less related to the number of topics, thereby improving the efficiency of event integration.

[0166] In this embodiment of the application, S3014 to S3016 are included before S3011; that is, before the event integration device obtains the matching result of the topic key string corresponding to each topic and the event to be integrated, the event integration method further includes S3014 to S3016. Each step is described below.

[0167] S3014. In at least one topic event corresponding to each topic in the topic library, obtain the topic event key string corresponding to each topic event.

[0168] It should be noted that the topic key string corresponding to a topic is obtained through the key string of at least one topic event. Here, the event integration device first obtains the topic event key string corresponding to each topic event. Since each topic includes at least one topic event, at least one topic event corresponds to at least one topic event key string.

[0169] S3015. Count the number of topic events corresponding to each sub-topic event key string in the topic event key string.

[0170] It should be noted that the event integration device counts the number of topic events corresponding to each sub-topic event key string in each topic event key string within at least one topic event key string, thereby obtaining the number of topic events corresponding to each sub-topic event key string, and thus obtaining the number of multiple topic events corresponding to multiple sub-topic event key strings under a topic.

[0171] S3016. Combine the sub-topic event key strings of the fourth set number of maximum number of topic events into the topic key string corresponding to each topic.

[0172] It should be noted that the event integration device selects the fourth set number of sub-topic event key strings corresponding to the number of multiple topic events under a topic, and determines the fourth set number of sub-topic event key strings as the topic key strings.

[0173] In this embodiment of the application, the event integration device in S3014 obtains the key string of the topic event corresponding to each topic event, which can be achieved through S30141 to S30143. The steps are described below.

[0174] S30141. Perform entity recognition for each topic event to obtain the entity key string corresponding to the preset entity type.

[0175] It should be noted that the event integration device obtains key strings of topic events from multiple dimensions; one of these dimensions is the entity of the topic event. The event integration device can obtain preset entity types in advance, such as person name type and place name type. Here, the event integration device performs entity recognition for each topic event, selects entities of preset entity types from the recognized entities, and uses the selected entities of preset entity types as entity key strings.

[0176] S30142. Perform string weight analysis on each topic event to obtain the key action strings.

[0177] It should be noted that the event integration device can also obtain key strings of topic events from the dimension of string weight; here, the event integration device analyzes the weight of strings in the topic event to obtain strings that are greater than the weight threshold, and determines the strings representing actions among the strings that are greater than the weight threshold as action key strings.

[0178] S30143. Determine the topic event key string based on one or both of the entity key string and the action key string.

[0179] It should be noted that when determining topic event keywords based on entity keywords, the event integration device can determine all entity keywords as topic event keywords, or it can extract strings from entity keywords to obtain topic event keywords. When determining topic event keywords based on action keywords, the event integration device can determine all action keywords as topic event keywords, or it can extract strings from action keywords to obtain topic event keywords. The event integration device can determine the string obtained from any combination of entity keywords and action keywords as the topic event keyword.

[0180] Understandably, since an event typically includes at least one of a person, a place, and an action, event integration devices can improve the accuracy of key strings in a topic event by determining the key strings of that topic event based on the strings associated with the person, place, and action respectively.

[0181] In this embodiment of the application, S30143 can be implemented by S301431 to S301433; that is, the event integration device determines the topic event key string based on one or both of the entity key string and the action key string, including S301431 to S301433. The steps are described below.

[0182] S301431. Obtain the number of entity key strings corresponding to entity key strings.

[0183] It should be noted that the event integration device can first determine the topic event key string based on the entity key string; here, when there is a limit to the number of strings in the key string of each topic event, the event integration device can determine how many strings to select from the action key string to be determined as the topic event key string based on the number of strings included in the entity key string, and can also determine whether to include the action key string as the topic event key string based on the number of strings included in the entity key string.

[0184] S301432. When the number of entity key strings is less than the fifth set number, the entity key strings and action key strings are combined into topic event key strings.

[0185] It should be noted that when there is a limit to the number of strings in the key strings of each topic event, and the number is set to the fifth limit, if the number of entity key strings is less than the fifth limit, the event integration device determines that the entity key strings are insufficient to serve as topic event key strings, and the action key strings need to be determined as topic event key strings as well; that is to say, at this time, the topic event key strings include entity key strings and action key strings.

[0186] S301433. When the number of entity key strings is greater than or equal to the fifth set quantity, the entity key string will be determined as the topic event key string.

[0187] It should be noted that when there is a limit to the number of strings in the key strings of each topic event, and the limit is set to the fifth number, the event integration device determines that there are enough strings in the entity key strings to serve as the key strings of the topic event when the number of entity key strings is greater than or equal to the fifth set number. In this case, the key strings of the topic event include the entity key strings.

[0188] See Figure 5 , Figure 5 This is another optional flowchart illustrating the event integration method provided in the embodiments of this application; as shown below. Figure 5 As shown in the embodiment of this application, S304 is followed by S308 to S311; that is, after the event integration device integrates the event to be integrated into the target topic and obtains an event context including the event to be integrated and at least one topic event, the event integration method further includes S308 to S311. Each step is described below.

[0189] S308, Present the search control.

[0190] It should be noted that the search control is used for information retrieval, and therefore, the search control can be used for searching for topics and events.

[0191] S309. In response to the first search operation performed on the search control, a simplified event context corresponding to the event context and a presentation control corresponding to the simplified event context are presented.

[0192] In this embodiment, when a user triggers the search control to search for information, if the searched information is related to the integrated target topic, the event integration device receives a first search operation applied to the search control; thus, the event integration device responds to the first search operation and presents the search results. Here, the presented search results may include a simplified event context corresponding to the event context, and a presentation control corresponding to the simplified event context.

[0193] It should be noted that the simplified event timeline is part of the event timeline and is presented as a subset of events within the event timeline; presentation controls are used to present the entire event timeline, such as the "View More" button, expand icons, etc.

[0194] S310. In response to a presentation operation performed on a presentation control, present an event pipeline, wherein each event in the presented event pipeline includes an event title and an event time, and the event is any one of the event to be integrated and at least one topic event.

[0195] It should be noted that when a user triggers the presentation control to view the entire event timeline, the event integration device receives the presentation operation applied to the presentation control. At this time, the event integration device responds to the presentation operation and presents the entire event timeline. Furthermore, the event integration device presents the event timeline by presenting the event title and event time of each event in the event timeline, where an event is any one of the events to be integrated and at least one topic event.

[0196] In this embodiment, the presented search results may include search recommendation results, which refer to recommended information for the integrated target topic, such as "Are you searching for 'the title of the integrated target topic'?". Here, when a user triggers an operation based on the search recommendation results, the event integration device can present a simplified event context corresponding to the event context, as well as a presentation control corresponding to the simplified event context, and present the event context in response to the presentation operation applied to the presentation control; or it can directly present the event context; this embodiment does not limit this.

[0197] For example, see Figure 6 , Figure 6 This is an exemplary event timeline diagram provided in an embodiment of this application; as shown... Figure 6 As shown, page 6-1 is the search results presentation page, which presents a simplified event timeline 6-11 corresponding to the event timeline, as well as a presentation control 6-12. When the presentation control 6-12 is clicked (presentation operation), the entire event timeline 6-21 is presented as shown in area 6-2. Here, each event in the presented event timeline is presented by presenting the event title (e.g., event title 6-211) and the event time (e.g., event time 6-212). Clicking the event title 6-211 presents the detailed information of the corresponding event.

[0198] For example, see Figure 7 , Figure 7 This is another exemplary event timeline presentation diagram provided in the embodiments of this application; as shown Figure 7 As shown, page 7-1 displays the search results, including other results and recommended search result 7-11. Clicking on recommended search result 7-11 will then display... Figure 6 The event timeline shown in the middle area 6-2 is 6-21.

[0199] S311. In response to a view operation applied to the event title or event time, display event details.

[0200] It should be noted that the event title or event time is a triggerable control, or each event has a corresponding control for viewing details. When a user triggers the event title, event time, or the control for viewing details, the event integration device receives the viewing operation applied to the event title or event time, or the viewing operation applied to the control for viewing details. At this time, the event integration device responds to the viewing operation and presents the event details information, which refers to the detailed description information of the event in the event timeline.

[0201] See Figure 8 , Figure 8This is another optional flowchart illustrating the event integration method provided in the embodiments of this application; as shown below. Figure 8 As shown in the embodiment of this application, S304 is followed by S312 to S314; that is, after the event integration device integrates the event to be integrated into the target topic and obtains an event context including the event to be integrated and at least one topic event, the event integration method further includes S312 to S314. Each step is described below.

[0202] S312, Present the last information to be presented for the target event.

[0203] It should be noted that the target event is any event among the events to be integrated and at least one topic event included in the event timeline; the final information to be presented refers to the information on the final presentation progress of the target event, such as the last page of the target event or the end of the target event.

[0204] S313. In the recommendation area corresponding to the last information to be presented, present the remaining events in the event hierarchy associated with the target event.

[0205] It should be noted that the page displaying the final information to be presented also includes a recommendation area, which is used to present recommended information. Here, the recommended information presented by the event integration device in the recommendation area consists of remaining events. Remaining events can be any event in the event timeline other than the target event, or they can be the latest development event in the event timeline other than the target event. These remaining events can be displayed through search results in the search box, links, etc., and this embodiment does not limit the scope of the application.

[0206] S314. In response to the second search operation on the remaining events, present detailed information about the remaining events.

[0207] It should be noted that when a user triggers a view operation for the remaining events, the event integration device also receives a second search operation for the remaining events; at this time, the event integration device responds to the second search operation by presenting detailed information about the remaining events to complete the response to the second search operation.

[0208] In the embodiments of this application, when the event integration device is implemented as a server, S308 to S314 can be implemented by the server; or the server can send the event context to the terminal and the terminal can implement it; the embodiments of this application do not limit this.

[0209] Understandably, event context can provide additional information beyond search terms, proactively uncovering relevant reading needs while satisfying search requirements, improving the completeness of information presentation on search results pages, reducing the number of searches that fail to obtain the target information, thereby reducing resource consumption in the search process, improving search conversion rates, and increasing user search frequency.

[0210] The following will describe an exemplary application of the embodiments of this application in a real-world application scenario.

[0211] It should be noted that for news topics with a long duration (corresponding to the aforementioned topics, which often consist of multiple events that have occurred (at least one topic event)), when the latest development event (the event to be integrated) of the news topic is obtained, the latest development event is attached to the corresponding news topic (target topic), forming an event timeline containing the latest development event; through the event timeline information, the development process of the event can be presented intuitively. When using the event integration method provided in the embodiments of this application to attach the latest development event to the corresponding information topic, it can be achieved through two stages: recall and classification, including the following steps:

[0212] First, based on the content of the latest developments, retrieve potentially relevant news topics (at least one topic to be integrated) from the news topic database (topic library).

[0213] It should be noted that each news topic in the news topic database corresponds to a topic keyword (topic key string). When the server matches any keyword (subtopic key string) in the latest development event with any of the topic keywords, it determines that the news topic is one of the potentially related news topics.

[0214] For example, see Figure 9 , Figure 9 This is an exemplary schematic diagram of news topic recall provided in an embodiment of this application; such as Figure 9As shown, "The First Department Responds to Zhang Er's Withdrawal of the Injunction Against the First Target" is the title of the latest development event 9-1. In the news topic database 9-2, news topic 9-21 includes 3 events, with corresponding topic keywords 9-211 being "nurse" and "vice president"; news topic 9-22 includes 4 events, with corresponding topic keywords 9-221 being "H location" and "jumping from the car"; news topic 9-23 includes 4 events, with corresponding topic keywords 9-231 being "Li San" and "first target". When the topic keywords of each news topic in the news topic database 9-2 are matched in the latest development event 9-1, since the topic keyword 9-231 corresponding to news topic 9-23, "first target", matches "first target" in the title of the latest development event 9-1, news topic 9-23 is one of the potentially related news topics to be recalled.

[0215] It should also be noted that topic keywords are the two keywords with the highest number of corresponding topic events among all keywords under a news topic (first quantity threshold). The topic keywords for each topic event can be obtained through entity recognition and word weight analysis (string weight analysis). Here, the server can use an entity recognition model (Char-WordUnion CNN, CWCNN) to perform entity recognition, and use the identified entities of the person name and place name type as the first keyword (entity key string). The server can use the "XGboost" model to perform word weight analysis, and use verbs among words with weights higher than the weight threshold as the second keyword (action key string). If the number of words with the first keyword is greater than 3 (fifth set quantity), the second keyword is no longer considered, and only the first keyword is used as the keyword of the topic event; if the number of words with the first keyword is less than 3, the first and second keywords are used together as the topic keywords of the topic event.

[0216] Then, the similarity between each potentially related news topic and the latest development event is obtained, so as to determine whether each news topic is related to the latest development event based on the similarity.

[0217] See Figure 10 , Figure 10 This is an exemplary diagram illustrating how to determine whether a news topic is related to the latest developments, provided in an embodiment of this application; for example... Figure 10As shown, the server obtains the similarity between each potentially relevant news topic and the latest development event from three aspects: vector semantic similarity 10-1 (semantic similarity), keyword graph similarity 10-2 (string graph similarity), and question-answer semantic similarity 10-3 (question-answer similarity). Finally, a fusion model 10-4 (e.g., the "XGboost" model, the "GBDT" model) is used to combine the vector semantic similarity 10-1, keyword graph similarity 10-2, and question-answer semantic similarity 10-3 to determine the decision score 10-5. Based on the decision score, it is determined whether each news topic is relevant to the latest development event, and finally, the relevant news topic 10-6 is obtained, which is the news topic to which the latest development event belongs.

[0218] The following explains the calculation process for the similarity of each dimension. When obtaining the vector semantic similarity 10-1, this includes vector semantic statistical similarity (semantic statistical similarity) and vector semantic self-attention similarity (semantic self-attention similarity); the acquisition of vector semantic statistical similarity will be explained first. The server calculates the similarity between the title of each topic event in the news topic and the title of the latest development event, obtaining the title vector semantic similarity (first sub-semantic similarity), and obtains the average title semantic similarity (average first sub-semantic similarity) and the maximum title semantic similarity (maximum first sub-semantic similarity). It also calculates the similarity between the keywords (key strings of the topic event) of each topic event in the news topic and the keywords (key strings of the event to be integrated) of the latest development event, obtaining the event keyword vector semantic similarity (second sub-semantic similarity), and obtains the average event keyword semantic similarity (average second sub-semantic similarity) and the maximum event keyword semantic similarity (maximum second sub-semantic similarity). Finally, it calculates the similarity between the topic keywords of the news topic and the keywords of the latest development event, obtaining the topic keyword vector semantic similarity (third sub-semantic similarity). Here, the title vector semantic similarity, average title semantic similarity, maximum title semantic similarity, event keyword vector semantic similarity, average event keyword semantic similarity, maximum event keyword semantic similarity, and topic keyword vector semantic similarity are collectively referred to as vector semantic statistical similarity.

[0219] It should be noted that calculating the similarity between the title of each topic event in a news topic and the title of the latest development event, calculating the similarity between the keywords of each topic event in a news topic and the keywords of the latest development event, and calculating the similarity between the topic keywords of a news topic and the keywords of the latest development event can all be achieved through a network model (semantic statistical similarity model).

[0220] See Figure 11a , Figure 11aThis is a schematic diagram of an exemplary model for obtaining vector semantic similarity provided in an embodiment of this application; as shown... Figure 11a As shown, network model 11-1 is used to obtain the similarity between two text pairs in terms of vector semantics, and network model 11-1 has a dual-tower structure. Here, the process of obtaining the semantic similarity of topic keyword vectors is used to explain the processing of network model 11-1: The topic keyword 11-2 corresponding to the news topic is input into the first network branch 11-11 (first semantic branch) of network model 11-1 to obtain the semantic vector 11-3 corresponding to the topic keyword 11-2; the keyword 11-4 corresponding to the latest development event is input into the second network branch 11-12 (second semantic branch) of network model 11-1 to obtain the semantic vector 11-5 corresponding to the keyword 11-4; then, the similarity between semantic vector 11-3 and semantic vector 11-5 is obtained through cosine similarity, thus obtaining the semantic similarity of topic keyword vectors 11-6. In addition, the first and second network branches can be the same network branches, such as both being "Bert" models; and semantic vectors 11-3 and 11-5 are both "cls" vectors of the first dimension output by each network branch, with the corresponding dimension being, for example, 768 dimensions; and during the training process of network model 11-1, it can be trained using 10,000 labeled sample pairs and a cross loss function.

[0221] When obtaining vector semantic self-attention similarity, see [link / reference]. Figure 11b , Figure 11b This is a schematic diagram of another exemplary model for obtaining vector semantic similarity provided in the embodiments of this application; as shown... Figure 11b As shown, the encoding module 11-71 in network model 11-7 (e.g., the "Bert" model) is used to obtain the semantic vector of each event in the event sequence consisting of the latest development event and at least one topic event 11-72 in a topic, resulting in a vector feature sequence 11-73 corresponding to the event sequence. Here, the server determines the segmentation identifier of the latest development event as 0 (distinguishing identifier) ​​and the segmentation identifier of the topic event as 1 (distinguishing identifier). The server obtains the semantic vector corresponding to 0 and the semantic vector corresponding to 1, then merges the semantic vector corresponding to 0 with the semantic vector of the latest development event (semantic features to be integrated), and merges the semantic vector corresponding to 1 with the semantic vector of the topic event (semantic features of the topic event). Finally, all the merged results are input into the Transformer model 11-74, which can obtain the vector semantic self-attention similarity.

[0222] It should be noted that transformation models 11-74 are network models in natural language processing, and transformation models 11-74 can be composed of at least one (e.g., three) transformation models stacked together. Furthermore, transformation models 11-74 automatically determine which topic event the latest development event should focus on for matching by calculating the adaptive relationship between the latest development event and the topic event for each pair of events (two sequence units) in the event sequence.

[0223] When obtaining the similarity of keyword graphs to 10⁻², the server constructs a keyword graph for each news topic and a keyword graph for the latest development event. Then, it obtains the representations of the two keyword graphs through a graph convolutional network (GCN) and calculates the cosine distance between the representations of the two keyword graphs to obtain the similarity of the keyword graphs.

[0224] The keyword graph for each news topic is constructed as follows: each keyword corresponding to each topic event under each news topic is treated as a graph node. If two keywords corresponding to two graph nodes belong to the same topic event, an edge is built between the two graph nodes to finally obtain the keyword graph corresponding to each news topic.

[0225] For example, see Table 1, which includes a title column and a keyword column for the topic event, as shown below:

[0226] Table 1

[0227]

[0228] See the keyword graph constructed based on Table 1. Figure 12 , Figure 12 This is a schematic diagram of an exemplary keyword diagram provided in an embodiment of this application; such as Figure 12 As shown, keywords Figure 12-1 The graph nodes are determined based on the keywords corresponding to each topic event in Table 1, including: First Object, Li San, Management, Blockade, First Organization, First Department, Prosecution, Plan, Second Department, Third Department, Prohibition, Third Organization, Second Object, Zhang Er, and Third Object; the edges between the graph nodes are as follows: Figure 12 As shown.

[0229] Similarly, see Figure 13 , Figure 13 This is a schematic diagram of another exemplary keyword diagram provided in the embodiments of this application; such as Figure 13 As shown, keywords Figure 13-1 The graph nodes are keywords of the latest progress event: Department 1, Zhang Er, and Object 1; and any two of Department 1, Zhang Er, and Object 1 have an edge.

[0230] When obtaining the semantic similarity of 10⁻³ between questions and answers, the server constructs a sequence of statements (the sequence of statements to be answered) based on the title and keywords of the information topic and the latest development events; and obtains the output of the statement sequence through the “MRC-BERT” model, and determines the semantic similarity between questions and answers based on the first dimension feature of the output.

[0231] See Figure 14 , Figure 14 This is an exemplary schematic diagram illustrating the acquisition of semantic similarity in question-and-answer processing, provided by an embodiment of this application; as shown... Figure 14 As shown, the statement sequence 14-1 is input into the network model 14-2 (e.g., the "MRC-BERT" model) to obtain the answer information 14-3. Then, based on the answer information 14-3, the semantic similarity 10-3 of the question-and-answer semantics is determined. In statement sequence 14-1, "CLS" indicates the beginning of the statement sequence, and "SEP" indicates the segmentation between statements. Statement sequence 14-1 also includes questions constructed based on news topics and the latest development events. For the latest development event 9-1 "The First Department Responds to Zhang Er's Withdrawal of the Ban on the First Target," with keywords "first target" and "Li San," and the news topic 9-23 with the title "Li San Blocks the First Target," the constructed statement sequence is "[CLS] Is the next sentence a development on the topic of Li San blocking the first target with keywords "first target" and "Li San"? [SEP] The First Department Responds to Zhang Er's Withdrawal of the Ban on the First Target [SEP]," or it could be "[CLS] Does the event of the First Department Responding to Zhang Er's Withdrawal of the Ban on the First Target belong to the topic of Li San blocking the first target with keywords "first target" and "Li San"? [SEP]

[0232] The following describes exemplary applications of the event integration method provided in the embodiments of this application.

[0233] It should be noted that network models 11-1, 11-7, the network model used to obtain string graph similarity, and 14-2 can be trained on more than 2000 topics during model training, constructing 50,000 topic-event sample pairs. For example, from all topics that have been launched and operated since a certain point in time, approved launched (corresponding to similarity) and unlaunched (corresponding to dissimilarity) events can be selected, with launched events as positive samples and unlaunched events as negative samples, constructing topic-event sample pairs according to the event order. For example, topic A contains five launched events (abcde) and two unlaunched events (fg) (where events de and fg both occur after events abc), thus constructing 6 topic-event sample pairs:

[0234] Positive samples: abc->d, abcd->e;

[0235] Negative samples: abc->f, abc->g, abcd->f, abcd->g.

[0236] See Figure 15 , Figure 15 This is a schematic diagram illustrating an exemplary feature importance provided in an embodiment of this application; as shown below. Figure 15 As shown, the vertical axis represents the importance index, which, in descending order of importance, is as follows: question-and-answer semantic similarity (10⁻³), vector semantic self-attention similarity (15⁻¹), maximum title semantic similarity (15⁻²), keyword graph similarity (10⁻²), maximum event keyword semantic similarity (15⁻³), average title semantic similarity (15⁻⁴), and topic keyword vector semantic similarity (15⁻⁵). Among these, the maximum title semantic similarity (15⁻²), maximum event keyword semantic similarity (15⁻³), average event keyword semantic similarity, average title semantic similarity (15⁻⁴), and topic keyword vector semantic similarity (15⁻⁵) together constitute... Figure 10 Vector semantic similarity in 10-1 is the statistical similarity of vector semantics.

[0237] Furthermore, when corrosion experiments were conducted on the similarity across eight dimensions (question-answer semantic similarity 10⁻³, vector semantic self-attention similarity 15⁻¹, maximum title semantic similarity 15⁻², keyword graph similarity 10⁻², maximum event keyword semantic similarity 15⁻³, average title semantic similarity 15⁻⁴, topic keyword vector semantic similarity 15⁻⁵, and average event keyword semantic similarity), the experimental results are shown in Table 2.

[0238] Table 2

[0239]

[0240] As shown in Table 2, when using 8 dimensions of similarity for event integration, the corresponding "AUC" is 0.9420. When the average title semantic similarity (15-4) is removed, the "AUC" decreases by -0.0025; when the maximum title semantic similarity (15-2) is removed, the "AUC" decreases by -0.0101; when the average event keyword semantic similarity is removed, the "AUC" decreases by -0.0043; when the maximum event keyword semantic similarity (15-3) is removed, the "AUC" decreases by -0.0000; and when the... When the semantic similarity of the keyword vectors is 15-5, the AUC decreases by -0.0013. When the semantic self-attention similarity of the vectors is removed (15-1), the AUC decreases by -0.0098. When the similarity of the keyword graphs is removed (10-2), the AUC decreases by -0.0081. When the semantic similarity of the question and answer is removed (10-3), the AUC decreases by -0.0152. Therefore, it is shown that the similarity of all eight dimensions contributes to the determination of the results when integrating events, which is consistent with the results corresponding to the importance index.

[0241] It should be noted that "AUC" refers to the area under the ROC curve (Receiver Operating Characteristic Curve) and the coordinate axis, and is a performance indicator.

[0242] In this embodiment, selection can also be made based on the accuracy and time consumption of each similarity, to determine whether the latest development event matches the topic event. See Table 3, which describes the time consumption of network model 11-1, network model 11-7, network model 14-2 used to obtain string graph similarity.

[0243] Table 3

[0244]

[0245] As shown in Table 3, the descending sequence based on time consumption is: network model 11-1, network model for obtaining string graph similarity, network model 11-7, and network model 14-2. Furthermore, since the semantic similarity of question and answer (10-3) is obtained through network model 14-2, the semantic self-attention similarity of vectors (15-1) is obtained through network model 11-7, the maximum semantic similarity of titles (15-2), the maximum semantic similarity of event keywords (15-3), the average semantic similarity of titles (15-4), the semantic similarity of topic keyword vectors (15-5), and the average semantic similarity of event keywords are obtained through network model 11-1, and the similarity of keyword graphs (10-2) is obtained through the network model used to obtain string graph similarity; therefore, the server uses the fastest and most accurate network models 11-7 and 14-2 as the initial calculation scheme. When the similarity obtained by network models 11-7 and 14-2 is high enough, a match is directly determined; when it is low enough, a mismatch is determined. Only when the similarity is in the middle range is network model 11-1 and the network model used to obtain string graph similarity used to calculate the similarity again, and all the obtained similarity combination features are input into the fusion model for final judgment.

[0246] For example, the determination process is shown in equation (1):

[0247] (1)

[0248] in, The similarity output by network model 11-7 This represents the similarity score output by network model 14-2.

[0249] It is understood that the embodiments of this application, by employing multi-dimensional heterogeneous features, integrate information from multiple events within a topic, and utilize three heterogeneous feature models: vector semantic features, keyword graph-based features, and question-answering semantic features, significantly enhancing the accuracy and rationality of similarity calculation. Furthermore, the event integration method provided in these embodiments enables automatic batch event integration without manual intervention, thereby improving event integration efficiency.

[0250] The following description continues to illustrate the exemplary structure of the event integration device 455 provided in the embodiments of this application as a software module. In some embodiments, such as... Figure 2 As shown, the software modules stored in the event integration device 455 of the memory 450 may include:

[0251] The information acquisition module 4551 is used to acquire events to be integrated and to acquire at least one topic to be integrated, wherein each topic to be integrated includes at least one topic event;

[0252] The similarity acquisition module 4552 is used to select one or more from semantic similarity, string graph similarity and question-answer similarity based on selection logic to obtain the target similarity between the event to be integrated and each topic to be integrated. The semantic similarity refers to the similarity in terms of semantic features, the string graph similarity refers to the similarity in terms of graph features corresponding to key strings, and the question-answer similarity refers to the similarity in terms of question-answer features.

[0253] The topic determination module 4553 is used to determine the target topic to which the event to be integrated belongs from at least one of the topics to be integrated, based on the target similarity.

[0254] The event integration module 4554 is used to integrate the events to be integrated into the target topic to obtain an event context including the events to be integrated and at least one topic event.

[0255] In this embodiment of the application, the selection logic includes one or more of the following: selection order, acquisition speed, accuracy, topic size, number of selections, topic type, model training scale, model applicability scope, and model applicability scale. The selection order is determined based on the priority of similarity. The acquisition speed is the speed at which similarity is acquired. The accuracy is the degree of accuracy of the similarity. The topic type is the content format of the topic to be integrated. The topic size is the size of at least one topic to be integrated. The model training scale is the training data size corresponding to the network model used to acquire each type of similarity.

[0256] In this embodiment of the application, when the selection logic includes the selection order, the similarity acquisition module 4552 is further configured to, based on the selection order, sequentially select a first predetermined number of similarities from the descending order of priority of the semantic similarity, the string graph similarity, and the question-and-answer similarity; obtain a comparison result between the first predetermined number of similarities and a similarity threshold; when the comparison result is a similarity result between the event to be integrated and the topic to be integrated, determine the first predetermined number of similarities as the target similarity between the event to be integrated and the topic to be integrated; when the comparison result is an undetermined similarity result between the event to be integrated and the topic to be integrated, continue to select the remaining similarities based on the selection order until the similarity result is determined or three of the semantic similarity, the string graph similarity, and the question-and-answer similarity are selected, and determine the selected multiple similarities as the target similarity between the event to be integrated and the topic to be integrated, wherein the remaining similarities are the similarities other than the first predetermined number of similarities among the semantic similarity, the string graph similarity, and the question-and-answer similarity.

[0257] In this embodiment of the application, when the selection logic includes the acquisition speed and the topic size, the similarity acquisition module 4552 is further configured to, when the topic size is greater than a set size, sequentially select a second set number of similarities from the descending order of the acquisition speed of the semantic similarity, the string graph similarity, and the question-answer similarity to obtain the target similarity between the event to be integrated and the topic to be integrated; when the topic size is less than or equal to the set size, sequentially select a third set number of similarities from the descending order of the acquisition speed of the semantic similarity, the string graph similarity, and the question-answer similarity to obtain the target similarity between the event to be integrated and the topic to be integrated, wherein the second set number is less than the third set number.

[0258] In this embodiment of the application, when the target similarity includes multiple types of semantic similarity, string graph similarity, and question-answering similarity, the topic determination module 4553 is further configured to determine the weight ratio of various similarities in the target similarity based on accuracy; based on the weight ratio, fuse the multiple similarities in the target similarity to obtain a discriminative similarity; and select the topic to be integrated corresponding to the highest discriminative similarity from at least one topic to be integrated to obtain the target topic to which the event to be integrated belongs.

[0259] In this embodiment, the semantic similarity includes semantic self-attention similarity. The similarity acquisition module 4552 is further configured to acquire the semantic features to be integrated corresponding to the event to be integrated, and the semantic features of each topic event in the topic to be integrated; enhance the semantic features to be integrated based on the difference identifiers of the event to be integrated and the topic events to obtain a first enhanced semantic feature, and enhance the semantic features of the topic events based on the difference identifiers to obtain a second enhanced semantic feature; form a semantic feature sequence by combining the first enhanced semantic feature with at least one second enhanced semantic feature corresponding to the topic to be integrated, and determine the semantic self-attention similarity based on the self-attention information between two sequence units in the semantic feature sequence.

[0260] In this embodiment of the application, the similarity acquisition module 4552 is further configured to: acquire a first sub-semantic similarity between the title of each topic event and the title of the event to be integrated in each topic to be integrated; and determine an average first sub-semantic similarity and a maximum first sub-semantic similarity based on the first sub-semantic similarity; acquire a second sub-semantic similarity between the topic event key string corresponding to each topic event and the event key string corresponding to the event to be integrated in each topic to be integrated; and determine an average second sub-semantic similarity and a maximum second sub-semantic similarity based on the second sub-semantic similarity; acquire a third sub-semantic similarity between the topic key string corresponding to each topic to be integrated and the event key string to be integrated; and determine the average first sub-semantic similarity, the maximum first sub-semantic similarity, the average second sub-semantic similarity, the maximum second sub-semantic similarity, and the third sub-semantic similarity as the semantic statistical similarity between the event to be integrated and each topic to be integrated.

[0261] In this embodiment, the first sub-semantic similarity, the second sub-semantic similarity, and the third sub-semantic similarity are obtained through a semantic statistical similarity model. The event integration device 455 further includes a model training module 4555 for acquiring training samples, wherein the training samples include a first string sample, a second string sample, and labeled similarity. Using the first semantic branch in the semantic statistical similarity model to be trained, a first estimated semantic corresponding to the first string sample is obtained; using the second semantic branch in the semantic statistical similarity model to be trained, a second estimated semantic corresponding to the second string sample is obtained; and based on the comparison result between the first estimated semantic and the second estimated semantic, the estimated similarity between the first string sample and the second string sample is determined. Based on the difference between the estimated similarity and the labeled similarity, backpropagation is performed in the semantic statistical similarity model to be trained to obtain the semantic statistical similarity model.

[0262] In this embodiment of the application, the similarity acquisition module 4552 is further configured to: determine each sub-topic event key string corresponding to at least one topic event as a graph node in each topic to be integrated; construct an edge between two graph nodes corresponding to two sub-topic event key strings belonging to the same topic event to obtain a first key string graph; construct a second key string graph based on the key string of the event to be integrated corresponding to the event to be integrated; and determine the string graph similarity between the event to be integrated and each topic to be integrated based on the comparison result between the vector representation of the first key string graph and the vector representation of the second key string graph.

[0263] In this embodiment of the application, the similarity acquisition module 4552 is further configured to combine a sequence of statements to be answered based on the title, topic key string and the event to be integrated for each of the topics to be integrated; obtain answer information of the sequence of statements to be answered; and determine the question-answer similarity between the event to be integrated and each of the topics to be integrated based on the answer information.

[0264] In this embodiment of the application, the information acquisition module 4551 is further configured to acquire, in the topic library, the matching result of the topic key string corresponding to each topic and the event to be integrated, wherein the topic library includes multiple topics; based on the matching result, when it is determined that at least one sub-topic key string in the topic key string matches the event to be integrated, the topic corresponding to the matching result is determined as the topic to be integrated that matches the event to be integrated; and at least one topic to be integrated that matches the event to be integrated is acquired from the topic library.

[0265] In this embodiment of the application, the information acquisition module 4551 is further configured to acquire a topic event key string corresponding to each topic event in at least one topic event corresponding to each topic in the topic library; count the number of topic events corresponding to each sub-topic event key string in the topic event key string; and combine the sub-topic event key strings with a fourth set number of maximum topic event counts into a topic key string corresponding to each topic.

[0266] In this embodiment of the application, the information acquisition module 4551 is further configured to perform entity recognition on each topic event to obtain entity key strings corresponding to preset entity types; perform string weight analysis on each topic event to obtain action key strings; and determine the topic event key strings based on one or both of the entity key strings and the action key strings.

[0267] In this embodiment of the application, the information acquisition module 4551 is further configured to acquire the number of entity key strings corresponding to the entity key string; when the number of entity key strings is less than a fifth preset number, the entity key string and the action key string are combined into the topic event key string; when the entity key string is greater than or equal to the fifth preset number, the entity key string is determined as the topic event key string.

[0268] In this embodiment, the event integration device 455 further includes an event display module 4556, used to present a search control; in response to a first search operation applied to the search control, a simplified event context corresponding to the event context and a presentation control corresponding to the simplified event context are presented, wherein the simplified event context belongs to the event context, and the presentation control is used to present the event context; in response to a presentation operation applied to the presentation control, the event context is presented, wherein each event in the presented event context includes an event title and an event time, and the event is any one of the event to be integrated and at least one of the topic events; in response to a viewing operation applied to the event title or the event time, event details information are presented.

[0269] In this embodiment of the application, the event display module 4556 is further configured to present the last information to be presented of the target event, wherein the target event is any one of the events to be integrated and at least one of the topic events included in the event network; in the recommendation area corresponding to the last information to be presented, the remaining events in the event network associated with the target event are presented, wherein the remaining events are any events in the event network other than the target event; and in response to a second search operation on the remaining events, the detailed information of the remaining events is presented.

[0270] This application provides a computer program product or computer program that includes computer instructions stored in a computer-readable storage medium. A processor of a computer device (event integration device) reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the event integration method described above in this application.

[0271] This application provides a computer-readable storage medium storing executable instructions. When these executable instructions are executed by a processor, they cause the processor to execute the event integration method provided in this application, for example... Figure 3 The event integration method is shown.

[0272] In some embodiments of this application, the computer-readable storage medium may be a memory such as FRAM, ROM, PROM, EPROM, EEPROM, flash memory, magnetic surface memory, optical disk, or CD-ROM; or it may be a device that includes one or any combination of the above-mentioned memories.

[0273] In some embodiments of this application, executable instructions may take the form of programs, software, software modules, scripts, or code, written in any form of programming language (including compiled or interpreted languages, or declarative or procedural languages), and may be deployed in any form, including being deployed as a standalone program or as a module, component, subroutine, or other unit suitable for use in a computing environment.

[0274] As an example, executable instructions may, but do not necessarily, correspond to files in a file system. They may be stored as part of a file that holds other programs or data, for example, in one or more scripts in a Hyper Text Markup Language (HTML) document, in a single file dedicated to the program in question, or in multiple collaborating files (e.g., a file that stores one or more modules, subroutines, or code sections).

[0275] As an example, executable instructions can be deployed to execute on a single computer device (in which case, this single computer device is the event consolidation device), or to execute on multiple computer devices located at one location (in which case, multiple computer devices located at one location are the event consolidation device), or to execute on multiple computer devices distributed across multiple locations and interconnected via a communication network (in which case, multiple computer devices distributed across multiple locations and interconnected via a communication network are the event consolidation device).

[0276] It is understood that in the embodiments of this application, data related to events to be integrated, topic events, and topics to be integrated are involved. When the embodiments of this application are applied to specific products or technologies, the permission or consent of the information subject is required, and the collection, use and processing of related data must comply with relevant laws, regulations and standards, conform to the principles of legality, legitimacy and necessity, do not involve obtaining data types prohibited or restricted by laws and regulations, and will not hinder the normal operation of the target website.

[0277] In summary, through the embodiments of this application, when determining the target topic to which an event belongs in at least one topic to be integrated, the target topic is determined by judging the target similarity between the event to be integrated and each topic to be integrated. That is, the target topic is determined by directly comparing the target similarity between the event to be integrated and each topic to be integrated. Furthermore, since target similarity includes one or more of semantic similarity, string graph similarity, and question-answering similarity, the obtained target similarity can accurately determine whether each topic to be integrated is the target topic to which the event to be integrated belongs. Therefore, when integrating the event to be integrated into the target topic, the accuracy of event integration can be improved. In addition, by recalling the event before obtaining the similarity during the event integration process, the efficiency of event integration can be improved.

[0278] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Any modifications, equivalent substitutions, and improvements made within the spirit and scope of this application are included within the scope of protection of this application.

Claims

1. An event integration method, characterized in that, include: Obtain events to be integrated and at least one topic to be integrated, wherein each topic to be integrated includes at least one topic event; Based on selection logic, one or more of semantic similarity, string graph similarity, and question-answer similarity are selected to obtain the target similarity between the event to be integrated and each topic to be integrated. Semantic similarity refers to similarity in terms of semantic features; string graph similarity refers to similarity in terms of graph features corresponding to key strings; and question-answer similarity refers to similarity in terms of question-answer features. The selection logic includes one or more of the following: selection order, acquisition speed, accuracy, topic size, number of selections, topic type, model training scale, model applicability, and model applicability scale. The selection order is determined based on the priority of similarity; the acquisition speed is the speed at which similarity is acquired; the accuracy is the accuracy of the similarity; the topic type is the content format of the topic to be integrated; the topic size is the size of at least one topic to be integrated; the model training scale is the training data size corresponding to the network model used to acquire each type of similarity; and the question-answer similarity refers to the probability that the event to be integrated belongs to each topic to be integrated, determined based on the interaction between the event to be integrated and each topic to be integrated, in machine reading comprehension. The target similarity includes the string graph similarity, which is obtained through the following steps: In each topic to be integrated, the key strings of each sub-topic event corresponding to at least one topic event are determined as graph nodes, and edges are built between the two graph nodes corresponding to the key strings of two sub-topic events belonging to the same topic event to obtain a first key string graph; based on the key strings of the event to be integrated corresponding to the event to be integrated, a second key string graph is constructed; based on the comparison result between the vector representation of the first key string graph and the vector representation of the second key string graph, the string graph similarity between the event to be integrated and each topic to be integrated is determined; Based on the target similarity, determine the target topic to which the event to be integrated belongs from at least one of the topics to be integrated; The events to be integrated are integrated into the target topic to obtain an event context that includes the events to be integrated and at least one event of the topic.

2. The method according to claim 1, characterized in that, When the selection logic includes the selection order, the step of selecting one or more from semantic similarity, string graph similarity, and question-answering similarity based on the selection logic to obtain the target similarity between the event to be integrated and each topic to be integrated includes: Based on the selection order, a first predetermined number of similarities are selected sequentially from the descending order of priority of the semantic similarity, the string graph similarity, and the question-answering similarity; Obtain the comparison results of the first set number of similarities with the similarity threshold; When the comparison result is a similarity result between the event to be integrated and the topic to be integrated, the first set number of similarities is determined as the target similarity between the event to be integrated and the topic to be integrated. When the comparison result is an undetermined similarity result between the event to be integrated and the topic to be integrated, the remaining similarities are selected based on the selection order until the similarity result is determined or three of the semantic similarity, string graph similarity, and question-answer similarity are selected. Then, the selected multiple similarities are determined as the target similarity between the event to be integrated and the topic to be integrated. The remaining similarities are the similarities other than the first set number of similarities among the semantic similarity, string graph similarity, and question-answer similarity.

3. The method according to claim 1, characterized in that, When the selection logic includes the acquisition speed and the topic size, the step of selecting one or more from semantic similarity, string graph similarity, and question-answer similarity based on the selection logic to obtain the target similarity between the event to be integrated and each topic to be integrated includes: When the topic size is larger than a set size, a second set number of similarities are selected sequentially from the descending order of the acquisition speed of the semantic similarity, the string graph similarity, and the question-answer similarity to obtain the target similarity between the event to be integrated and the topic to be integrated. When the topic size is less than or equal to the set size, a third set number of similarities are selected sequentially from the descending order of the acquisition speed of the semantic similarity, the string graph similarity, and the question-answer similarity to obtain the target similarity between the event to be integrated and the topic to be integrated, wherein the second set number is less than the third set number.

4. The method according to any one of claims 1 to 3, characterized in that, When the target similarity includes multiple factors such as semantic similarity, string graph similarity, and question-answering similarity, determining the target topic to which the event to be integrated belongs from at least one of the topics to be integrated based on the target similarity includes: The weight ratio of various similarities in the target similarity is determined based on the accuracy. Based on the weighting ratio, multiple similarities in the target similarity are fused to obtain a discriminative similarity. From at least one of the topics to be integrated, select the topic corresponding to the highest discriminative similarity to obtain the target topic to which the event to be integrated belongs.

5. The method according to any one of claims 1 to 3, characterized in that, The semantic similarity includes semantic self-attention similarity, which is obtained through the following steps: Obtain the semantic features of the event to be integrated, and the semantic features of each topic event in the topic to be integrated; Based on the distinguishing identifiers of the event to be integrated and the topic event, the semantic features of the event to be integrated are enhanced to obtain a first enhanced semantic feature, and the semantic features of the topic event are enhanced based on the distinguishing identifiers to obtain a second enhanced semantic feature; The first enhanced semantic feature and at least one second enhanced semantic feature corresponding to the topic to be integrated are combined to form a semantic feature sequence, and the semantic self-attention similarity is determined based on the self-attention information between two sequence units in the semantic feature sequence.

6. The method according to any one of claims 1 to 3, characterized in that, The semantic similarity includes semantic statistical similarity, which is obtained through the following steps: In each of the topics to be integrated, the first sub-semantic similarity between the title of each topic event and the title of the event to be integrated is obtained, and based on the first sub-semantic similarity, the average first sub-semantic similarity and the maximum first sub-semantic similarity are determined. In each topic to be integrated, the second sub-semantic similarity between the topic event key string corresponding to each topic event and the event key string corresponding to the event to be integrated is obtained, and based on the second sub-semantic similarity, the average second sub-semantic similarity and the maximum second sub-semantic similarity are determined; Obtain the third sub-semantic similarity between the topic key string corresponding to each topic to be integrated and the event key string to be integrated; The average first sub-semantic similarity, the maximum first sub-semantic similarity, the average second sub-semantic similarity, the maximum second sub-semantic similarity, and the third sub-semantic similarity are determined as the semantic statistical similarity between the event to be integrated and each topic to be integrated.

7. The method according to any one of claims 1 to 3, characterized in that, The target similarity includes the question-and-answer similarity, which is obtained through the following steps: Based on the title, key strings of the topic, and the event to be integrated for each topic to be integrated, a sequence of statements to be answered is combined; Obtain the answer information for the sequence of statements to be answered; Based on the answer information, the question-and-answer similarity between the event to be integrated and each topic to be integrated is determined.

8. The method according to any one of claims 1 to 3, characterized in that, The acquisition of at least one topic to be integrated includes: In the topic library, the matching result between the topic key string corresponding to each topic and the event to be integrated is obtained, wherein the topic library includes multiple topics; Based on the matching result, when at least one sub-topic key string in the topic key string matches the event to be integrated, the topic corresponding to the matching result is determined as the topic to be integrated that matches the event to be integrated. Obtain at least one topic from the topic library that matches the event to be integrated.

9. The method according to claim 8, characterized in that, Before obtaining the matching results between the topic key string corresponding to each topic and the event to be integrated in the topic library, the method further includes: In at least one topic event corresponding to each topic in the topic library, entity recognition is performed on each topic event to obtain entity key strings corresponding to preset entity types; Perform string weight analysis on each of the aforementioned topic events to obtain key action strings; Determine the topic event key string based on one or both of the entity key string and the action key string; Count the number of topic events corresponding to each sub-topic event key string in the topic event key string; From the descending order of the number of topic events corresponding to the topic event key strings, a fourth predetermined number of sub-topic event key strings are selected sequentially to obtain the topic key string corresponding to each topic.

10. The method according to any one of claims 1 to 3, characterized in that, After integrating the events to be integrated into the target topic to obtain an event timeline including the events to be integrated and at least one event from the topic, the method further includes: Present the search control; In response to a first search operation performed on the search control, a simplified event context corresponding to the event context and a presentation control corresponding to the simplified event context are presented, wherein the simplified event context belongs to the event context and the presentation control is used to present the event context; In response to a presentation operation performed on the presentation control, the event hierarchy is presented, wherein each event in the presented event hierarchy includes an event title and an event time, and the event is any one of the event to be integrated and at least one of the topic events; In response to a view operation applied to the event title or the event time, event details are displayed.

11. The method according to any one of claims 1 to 3, characterized in that, The semantic similarity includes one or both of semantic self-attention similarity and semantic statistical similarity. The semantic self-attention similarity is determined based on the self-attention between the event to be integrated and the topic event. The semantic statistical similarity is obtained by comparing the vector semantic features of the title, key strings and body of the event to be integrated with the vector semantic features of the title, key strings and related information of each topic event to be integrated.

12. An event integration device, characterized in that, include: The information acquisition module is used to acquire events to be integrated and to acquire at least one topic to be integrated, wherein each topic to be integrated includes at least one topic event; The similarity acquisition module is used to select one or more from semantic similarity, string graph similarity, and question-answer similarity based on selection logic to obtain the target similarity between the event to be integrated and each topic to be integrated. Semantic similarity refers to similarity in terms of semantic features; string graph similarity refers to similarity in terms of graph features corresponding to key strings; and question-answer similarity refers to similarity in terms of question-answer features. The selection logic includes one or more of the following: selection order, acquisition speed, accuracy, topic size, number of selections, topic type, model training scale, model applicability scope, and model applicability scale. The order is determined based on the priority of similarity, the acquisition speed is the speed of acquiring similarity, the accuracy is the accuracy of similarity, the topic type is the content form of the topic to be integrated, the topic size is the size of at least one topic to be integrated, the model training size is the training data size corresponding to the network model used to acquire each type of similarity, and the question-answer similarity refers to the probability that the event to be integrated belongs to each topic to be integrated, based on the event to be integrated and each topic to be integrated, constructing questions and articles in machine reading comprehension, and determining the answer information through the interaction of questions and articles; The target similarity includes the string graph similarity. The similarity acquisition module is further configured to, in each of the topics to be integrated, determine each sub-topic event key string corresponding to at least one topic event as a graph node, and build edges between two graph nodes corresponding to two sub-topic event key strings belonging to the same topic event to obtain a first key string graph; construct a second key string graph based on the key string of the event to be integrated corresponding to the event to be integrated; and determine the string graph similarity between the event to be integrated and each of the topics to be integrated based on the comparison result between the vector representation of the first key string graph and the vector representation of the second key string graph. A topic determination module is used to determine the target topic to which the event to be integrated belongs from at least one of the topics to be integrated, based on the target similarity. The event integration module is used to integrate the events to be integrated into the target topic to obtain an event outline including the events to be integrated and at least one event of the topic.

13. The event integration device according to claim 12, characterized in that, The similarity acquisition module is also used to determine the weight ratio of various similarities in the target similarity based on the accuracy. Based on the weighting ratio, multiple similarities in the target similarity are fused to obtain a discriminative similarity. From at least one of the topics to be integrated, select the topic corresponding to the highest discriminative similarity to obtain the target topic to which the event to be integrated belongs.

14. The event integration device according to claim 12, characterized in that, The semantic similarity includes semantic statistical similarity. The similarity acquisition module is further configured to: acquire a first sub-semantic similarity between the title of each topic event and the title of the event to be integrated in each topic to be integrated; and determine an average first sub-semantic similarity and a maximum first sub-semantic similarity based on the first sub-semantic similarity; acquire a second sub-semantic similarity between the topic event key string corresponding to each topic event and the event key string corresponding to the event to be integrated in each topic to be integrated; and determine an average second sub-semantic similarity and a maximum second sub-semantic similarity based on the second sub-semantic similarity; acquire a third sub-semantic similarity between the topic key string corresponding to each topic to be integrated and the event key string to be integrated; and determine the average first sub-semantic similarity, the maximum first sub-semantic similarity, the average second sub-semantic similarity, the maximum second sub-semantic similarity, and the third sub-semantic similarity as the semantic statistical similarity between the event to be integrated and each topic to be integrated.

15. An event integration device, characterized in that, include: Memory, used to store executable instructions; A processor, when executing executable instructions stored in the memory, implements the event integration method according to any one of claims 1 to 11.

16. A computer-readable storage medium, characterized in that, It stores executable instructions for implementing the event integration method according to any one of claims 1 to 11 when executed by a processor.

17. A computer program product, characterized in that, The device stores executable instructions that, when executed by a processor, implement the event integration method according to any one of claims 1 to 11.

Images