Abstract information generation method and device

Abstract

The invention discloses a summary information generation method and device, and relates to the technical field of computers. A specific embodiment of the method comprises the steps of extracting sentence structure information of a to-be-processed text; and generating summary information of the to-be-processed text based on the sentence structure information and the semantic information of the to-be-processed text. According to the embodiment, the grammar rationality and semantic integrity of abstract generation are effectively improved.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a method and device for generating summary information. Background Art

[0002] The Internet is in a period of rapid development, and network information has also experienced explosive growth. As an important branch of this research field, automatic text summarization (abbreviated as automatic summarization) technology can extract short key information from massive texts to help people quickly understand the original text content, effectively saving people's time and energy in obtaining useful information.

[0003] At present, summary generation methods can be divided into extractive summary and abstract summary according to the implementation method. Among them, extractive summary extracts important sentences that already exist in the original text to form a summary, while abstract summary generates new sentences as a summary by understanding relevant semantic information and using natural language technology. However, the grammatical rationality and semantic integrity of the summaries generated by the two are poor. Summary of the invention

[0004] Embodiments of the present application provide a summary information generation method, apparatus, device, and storage medium.

[0005] According to a first aspect, an embodiment of the present application provides a summary information generation method, the method comprising: extracting sentence structure information of a text to be processed; and generating summary information of the text to be processed based on the sentence structure information and semantic information of the text to be processed.

[0006] According to a second aspect, an embodiment of the present application provides a summary information generating device, which includes: an extraction module and a generation module, wherein the extraction module is configured to extract sentence structure information of a text to be processed; and the generation module is configured to generate summary information of the text to be processed based on the sentence structure information and semantic information of the text to be processed.

[0007] According to the third aspect, an embodiment of the present application provides an electronic device, which includes one or more processors; a storage device on which one or more programs are stored, and when the one or more programs are executed by the one or more processors, the one or more processors implement the method described in the first aspect.

[0008] According to a fourth aspect, an embodiment of the present application provides a computer-readable medium having a computer program stored thereon, which, when executed by a processor, implements the method described in the first aspect.

[0009] According to a fifth aspect, an embodiment of the present disclosure provides a computer program product, including a computer program, which implements the method described in the embodiment of the first aspect when executed by a processor.

[0010] The present application generates summary information of the text to be processed by extracting sentence structure information of the text to be processed; based on the sentence structure information and the semantic information of the text to be processed, that is, for each sentence in the text to be processed, its corresponding semantics and structure are fully considered, and the two are encoded and integrated, which effectively improves the ability to obtain text information, and then improves the grammatical rationality and semantic integrity of the generated summary.

[0011] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present disclosure, nor is it intended to limit the scope of the present disclosure. Other features of the present disclosure will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 is an exemplary system architecture diagram to which the present application may be applied;

[0013] Figure 2 is a flowchart of an embodiment of a method for generating summary information according to the present application;

[0014] Figure 3 is a flowchart of another embodiment of a method for generating summary information according to the present application;

[0015] Figure 4 is a flow chart of an application scenario of the method for generating summary information according to the present application;

[0016] Figure 5 is a schematic diagram of an embodiment of a summary information generating device according to the present application;

[0017] Figure 6 It is a structural diagram of a computer system suitable for implementing a server of an embodiment of the present application. DETAILED DESCRIPTION

[0018] The following is a description of exemplary embodiments of the present application in conjunction with the accompanying drawings, including various details of the embodiments of the present application to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the present application. Similarly, for the sake of clarity and conciseness, the description of well-known functions and structures is omitted in the following description.

[0019] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0020] Figure 1An exemplary system architecture 100 is shown to which an embodiment of the summary information generation method of the present application can be applied.

[0021] like Figure 1 As shown, the system architecture 100 may include terminal devices 101, 102, 103, a network 104 and a server 105. The network 104 is used to provide a medium for communication links between the terminal devices 101, 102, 103 and the server 105. The network 104 may include various connection types, such as wired, wireless communication links or optical fiber cables, etc.

[0022] The terminal devices 101, 102, 103 interact with the server 105 via the network 104 to receive or send messages, etc. Various communication client applications may be installed on the terminal devices 101, 102, 103, such as summary information generation applications, communication applications, etc.

[0023] Terminal devices 101, 102, 103 can be hardware or software. When terminal devices 101, 102, 103 are hardware, they can be various electronic devices with display screens, including but not limited to mobile phones and laptops. When terminal devices 101, 102, 103 are software, they can be installed in the electronic devices listed above. It can be implemented as multiple software or software modules (for example, to provide summary information generation services), or it can be implemented as a single software or software module. No specific limitation is made here.

[0024] The server 105 may be a server that provides various services, for example, extracting sentence structure information of the text to be processed; generating summary information of the text to be processed based on the sentence structure information and semantic information of the text to be processed.

[0025] It should be noted that the server 105 can be hardware or software. When the server 105 is hardware, it can be implemented as a distributed server cluster consisting of multiple servers, or it can be implemented as a single server. When the server is software, it can be implemented as multiple software or software modules (for example, for providing summary information generation services), or it can be implemented as a single software or software module. No specific limitation is made here.

[0026] It should be noted that the summary information generation method provided in the embodiments of the present disclosure may be executed by the server 105, or by the terminal devices 101, 102, 103, or by the server 105 and the terminal devices 101, 102, 103 cooperating with each other. Accordingly, the various parts (such as various units, sub-units, modules, sub-modules) included in the summary information generation apparatus may be all arranged in the server 105, or all arranged in the terminal devices 101, 102, 103, or respectively arranged in the server 105 and the terminal devices 101, 102, 103.

[0027] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to the implementation requirements.

[0028] Figure 2 A flowchart 200 of an embodiment of a method for generating summary information that can be applied to the present application is shown. In this embodiment, the method for generating summary information includes the following steps:

[0029] Step 201: extract sentence structure information of the text to be processed.

[0030] In this embodiment, the execution subject (such as Figure 1 The server 105 or terminal devices 101, 102, 103 shown in the figure can obtain the text to be processed locally or from a remote server storing the text to be processed, perform sentence processing on the text to be processed, and extract sentence structure information of each sentence after the sentence processing.

[0031] Here, the text to be processed may be any text from which summary information can be extracted, such as news text, book text, etc. The text to be processed may include text content such as title and body.

[0032] Among them, the sentence structure information can be determined based on information such as the keywords of the sentence, the number of keywords, the location of keywords, the length of the sentence, the relevance of keywords, and the similarity with key sentences.

[0033] Here, the sentence structure information may include only the structure information within a sentence, only the structure information between sentences, or both the structure information within a sentence and the structure information between sentences, and this application does not limit this.

[0034] Among them, the structural information between sentences may include the position information of the sentences (such as at the beginning of a paragraph, at the end of a paragraph, etc.), the relationship between sentences (such as cause and effect, parallel relationship, general-specific relationship, hypothetical relationship, etc.), etc.

[0035] In some optional methods, sentence structure information of the text to be processed is extracted, including: segmenting the text to be processed, and determining the sentence structure information of the text to be processed based on at least one of the following attribute information of the sentence: keyword relevance, number of named entities, sentence length, number of summary keywords, and similarity with key sentences.

[0036] In this implementation, the execution entity may first perform sentence processing on the text to be processed, and then determine the sentence structure information of the text to be processed based on one or more of the following attribute information of the sentences after sentence processing: keyword relevance, number of named entities, sentence length, number of summary keywords, and similarity with key sentences.

[0037] Among them, the execution subject can use a graph-based sorting algorithm to determine the keywords of the text to be processed, and determine the keyword relevance based on the number of keywords contained in the current sentence. Usually, the keyword relevance is positively correlated with the number of keywords, that is, the more keywords the current sentence contains, the higher the keyword relevance, and the fewer keywords the current sentence contains, the lower the keyword relevance.

[0038] The execution entity can use a graph-based sorting algorithm to determine the key sentences in the text to be processed and adopt an algorithm for calculating similarity in existing technologies or future development technologies, such as the Jaccard similarity coefficient, cosine similarity, etc., to calculate the similarity between the current sentence and the key sentence; further, based on the similarity between the current sentence and the key sentence, determine the similarity between the current sentence and the key sentence.

[0039] Since a sentence usually contains more named entities, that is, entities identified by names (for example, names of people, names of organizations, names of places, etc.), the more important the sentence is, the execution subject can perform word segmentation on the current sentence and mark the parts of speech, and determine the number of named entities in the sentence based on the words and the marked parts of speech.

[0040] Since longer sentences contain more information and are more important, the execution subject can perform word segmentation on the current sentence and determine the length of the statistically segmented words as the sentence length.

[0041] Since if a sentence contains words such as "in short" and "it can be concluded that", that is, the sentence is a summary word, the importance of the sentence is higher. The execution entity can establish a preset summary keyword word list and determine whether the current sentence contains summary keywords and the number of summary keywords based on the summary keyword word list.

[0042] This implementation method divides the text to be processed into sentences and determines the sentence structure information of the text to be processed based on at least one of the following attribute information of the sentence: keyword relevance, number of named entities, sentence length, number of summary keywords, and similarity with key sentences, thereby improving the accuracy and reliability of the acquired sentence structure information.

[0043] Step 202: Generate summary information of the text to be processed based on the sentence structure information and the semantic information of the text to be processed.

[0044] In this embodiment, the execution subject may determine the semantic information of the text to be processed based on the words in the text to be processed, and generate summary information of the text to be processed according to the sentence structure information and the semantic information of the text to be processed.

[0045] Here, the execution subject can determine the summary information of the text to be processed based on the sentence structure information, the semantic information of the text to be processed and the preset mapping relationship table of sentence structure information, semantic information and summary information; or the sentence structure information and the semantic information of the text to be processed can be input into a preset generation network to generate the summary information of the text to be processed, which is not limited in this application.

[0046] In some optional methods, the semantic information of the text to be processed is obtained by: counting the frequency of words in the text to be processed and sorting them from large to small based on the frequency; determining the semantic information of the text to be processed based on a preset number of words ranked at the top.

[0047] In this implementation, the execution entity can directly count the frequency of words in the text to be processed and sort them in descending order based on the frequency; or it can first perform sentence processing on the text to be processed, and then perform word segmentation and stop word removal to obtain the processed text to be processed, count the frequency of words in the processed text to be processed and sort them in descending order based on the frequency. This application does not limit this.

[0048] Furthermore, the execution subject may determine the semantic information of the text to be processed based on a preset number of words ranked at the top.

[0049] Among them, the preset number can be set according to experience and actual needs, for example, 1,000, 200, etc., and this application does not limit this.

[0050] This implementation method counts the word frequencies of words in the text to be processed and sorts them from large to small based on the word frequencies; based on a preset number of words that are ranked first, the semantic information of the text to be processed is determined, thereby improving the accuracy of the determined semantic information.

[0051] Continue to refer Figure 3, which shows a process 300 of another embodiment of the summary information generation method of the present application. The summary information generation method includes the following steps:

[0052] Step 301: extract sentence structure information of the text to be processed.

[0053] In this embodiment, the implementation details and technical effects of step 301 can refer to the description of step 201 and will not be repeated here.

[0054] Step 302: input the sentence structure information and the semantic information of the text to be processed into the generation network to generate summary information of the text to be processed.

[0055] In this embodiment, the execution subject can directly input the sentence structure information and the semantic information of the text to be processed into the generation network to generate summary information of the text to be processed.

[0056] The generation network is used to represent the correspondence between sentence structure information, semantic information and summary information. The generation network can be trained based on any neural network for generating information (e.g., Seq2Seq network, pointer generation network, etc.) in existing technology or future development technology.

[0057] Specifically, the generation network may include an encoder and a decoder. The encoder and the decoder may be trained together or separately, and this application does not limit this.

[0058] In some optional methods, sentence structure information and semantic information of the text to be processed are input into a generation network to generate summary information of the text to be processed, including: inputting the semantic information into a first encoding network to generate semantic encoding information; inputting the semantic encoding information and text structure information into a second encoding network to generate fused encoding information; and inputting the fused encoding information into a decoding network to generate summary information of the text to be processed.

[0059] In this implementation, the generation network may include a first encoding network, a second encoding network and a decoding network. The execution subject may input semantic information into the first encoding network to generate semantic encoding information; input the semantic encoding information and text structure information into the second encoding network to generate fused encoding information; and input the fused encoding information into the decoding network to generate summary information of the text to be processed.

[0060] Among them, the first coding network and the second coding network can be any network for encoding information in the existing technology or future development technology, for example, RNN (Recurrent Neural Network), LSTM (Long Short Term Memory), etc. The first coding network and the second coding network can be the same or different, and this application does not limit this.

[0061] The decoding network can be any network for decoding information in the existing technology or future development technology, for example, a GRU (Gate Recurrent Unit) network, a LSTM (Long Short Term Memory) network, etc., and this application does not limit this.

[0062] Furthermore, the decoding network can be a decoding network integrated with an attention mechanism (i.e., Attention mechanism). The decoding network integrated with the attention mechanism will consider the output of the previous moment and the corresponding text semantic information and sentence structure information input at the current moment during decoding. The output information of the previous moment and the input information at this moment can be used as elements for weight setting to focus the decoding "attention" on important current information.

[0063] This implementation method generates semantic coding information by inputting semantic information into a first encoding network; inputting the semantic coding information and text structure information into a second encoding network to generate fused coding information; and inputting the fused coding information into a decoding network to generate summary information of the text to be processed, thereby achieving separate encoding and fusion of semantic information and sentence structure information, and improving the ability of the encoding network to obtain text information.

[0064] In some optional methods, semantic information is input into the first encoding network to generate semantic encoding information, including: inputting the semantic information into the first sub-network to generate forward semantic encoding information; inputting the semantic information into the second sub-network to generate reverse semantic encoding information; inputting the forward semantic encoding information and the reverse semantic encoding information into the fusion network to generate semantic encoding information.

[0065] In this implementation, the first coding network may include a first subnetwork, a second subnetwork and a fusion network. The execution entity may input semantic information into the first subnetwork to generate forward semantic coding information; input semantic information into the second subnetwork to generate reverse semantic coding information; and input forward semantic coding information and reverse semantic coding information into the fusion network to generate semantic coding information.

[0066] Among them, the first subnetwork and the second subnetwork can be any network for encoding information in the existing technology or future development technology, for example, RNN (Recurrent Neural Network), LSTM (Long Short Term Memory), etc. The first subnetwork and the second subnetwork can be the same or different, and this application does not limit this.

[0067] Here, the fusion network may fuse the forward semantic coding information and the reverse semantic coding information in a variety of ways, such as splicing, convolution, etc., wherein the splicing may include head-to-tail splicing, splicing into vectors, etc.

[0068] This implementation method generates forward semantic coding information by inputting semantic information into the first sub-network; generates reverse semantic coding information by inputting semantic information into the second sub-network; and generates semantic coding information by inputting the forward semantic coding information and the reverse semantic coding information into the fusion network, that is, bidirectionally encoding and fusing the input semantic information, thereby effectively improving the accuracy of the generated semantic coding information.

[0069] In some optional embodiments, the generation network is a pointer generation network integrated with a convergence mechanism.

[0070] In this implementation, the execution subject may input sentence structure information and semantic information of the text to be processed into the generation network to generate summary information of the text to be processed.

[0071] Among them, the generating network is a pointer generating network (i.e., Pointer-Generator Networks) integrated with a convergence mechanism (i.e., Coverage mechanism).

[0072] Here, the pointer generator network can copy words from the source text by pointing, which helps to accurately copy the information while retaining the generative ability of the generator; the convergence mechanism can track the content of the summary and continuously update the attention, thereby preventing the text from being repeated.

[0073] from Figure 3 It can be seen that Figure 2 Compared with the corresponding embodiments, the process 300 of the summary information generation method in this embodiment embodies extracting sentence structure information of the text to be processed, inputting the sentence structure information and the semantic information of the text to be processed into the generation network, and generating summary information of the text to be processed, that is, encoding and fusing the sentence structure information and the semantic information through the generation network to generate summary information, thereby effectively improving the accuracy of the generated summary information.

[0074] Continue to see Figure 4 , Figure 4 It is a scene diagram of the summary information generating method according to this embodiment.

[0075] exist Figure 4In the application scenario, the execution subject can obtain the text to be processed locally or from a remote server storing the text to be processed, such as a news text. Further, the execution subject can perform word segmentation on the text to be processed, count the frequency of the words in the text to be processed, and sort them from highest to lowest based on the frequency; based on a preset number of words ranked at the top, determine the semantic information of the text to be processed, such as a vocabulary determined based on a preset number of words ranked at the top. Further, the execution subject can input the semantic information into the first encoding network, such as a single-layer Bi-LSTM (Bidirectional LongShort Term Memory) to generate semantic encoding information; further, input the sentence structure information into the second encoding network, such as a single-layer LSTM, to obtain fused encoding information.

[0076] Furthermore, the fused encoded information is input into a decoding network integrated with an attention mechanism (i.e., attention mechanism), such as a single-layer LSTM, to obtain summary information of the text to be processed.

[0077] It should be noted that, in the process of generating summary information, the sentence structure information of each sentence of the text to be processed may correspond to the semantic information of the sentence.

[0078] In the training process of the first encoding network, the second encoding network and the decoding network, the first encoding network can generate a hidden state vector h according to the input semantic information. i The second encoding network can generate a hidden state vector based on the input sentence structure information, and convolve the hidden state vector to h at a preset ratio. i In the above equation, we get h i’ , at each time step t, the decoding network can retain the generated hidden layer state vector, such as, s t .s t It is used to understand the information of the output words, and use it as a basis to obtain the different effects of the input sequence on the next output sequence.

[0079] For the current time step, after calculating the attention distribution of the current step, the decoding network can output h of the second encoding network. i’ Do weighted averaging to obtain the dynamic representation of the original text, i.e., the context vector h t * , and at the same time, maintain a convergence mechanism vector c t , that is, the coverage vector, which is the cumulative sum of the attention distributions calculated at all past time steps and is used to record which words the model has paid attention to.

[0080] c t 、h t *It can be expressed by the following formula:

[0081]

[0082] a t =softmax(e t )

[0083]

[0084] Among them, v, W h , W s 、w c and b attn is the learning parameter, i represents the i-th word, represents the weight of the i-th word at the t-th time step, represents the attention distribution of the i-th word, a t represents the attention distribution of the whole text, represents the aggregation mechanism vector of the i-th word, c t Represents the full-text aggregation mechanism vector.

[0085] The final distribution prediction of word w is P(w), which is shown in the following formula:

[0086]

[0087] Among them, P vocab (w) is the probability distribution of the word predicted in the current step in the vocabulary, p gen The probability of generating each word calculated dynamically for each step is used to decide whether to generate a word from the vocabulary or copy the word from the source text, vector w s 、w x , b ptr , V, V, b and b' are learning parameters, σ is the sigmoid function, x t Embedded vector representing the current decoding step.

[0088] Here, during the training of the first encoding network, the second encoding network and the decoding network, the loss function loss t It can be expressed by the following formula:

[0089]

[0090] Among them, covloss t is the convergence loss function, namely coverage loss, which is used to avoid the attention mechanism from repeatedly focusing on the same location. λ is a hyperparameter. Represents the target word at time step t in the training phase The probability distribution of represents the target word at time step t The logarithmic loss of .

[0091] Further references Figure 5 As an implementation of the methods shown in the above figures, the present disclosure provides an embodiment of a summary information generating device. Figure 2 Corresponding to the embodiment of the summary information generating method shown, the device can be specifically applied to various electronic devices.

[0092] like Figure 5 As shown, the summary information generating device 500 of this embodiment includes: an extraction module 501 and a generation module 502 .

[0093] The extraction module 501 may be configured to extract sentence structure information of the text to be processed.

[0094] The generating module 502 may be configured to generate summary information of the text to be processed based on the sentence structure information and the semantic information of the text to be processed.

[0095] In some optional aspects of this embodiment, the generation module is further configured to input the sentence structure information and the semantic information of the text to be processed into the generation network to generate summary information of the text to be processed.

[0096] In some optional embodiments of this embodiment, the generation module is further configured to input semantic information into the first encoding network to generate semantic encoding information; input the semantic encoding information and sentence structure information into the second encoding network to generate fused encoding information; input the fused encoding information into the decoding network to generate summary information of the text to be processed.

[0097] In some optional methods of this embodiment, semantic information is input into the first encoding network to generate semantic coding information, including: inputting the semantic information into the first sub-network to generate forward semantic coding information; inputting the semantic information into the second sub-network to generate reverse semantic coding information; inputting the forward semantic coding information and the reverse semantic coding information into the fusion network to generate semantic coding information.

[0098] In some optional manners of this embodiment, the generation network is a pointer generation network integrated with a convergence mechanism.

[0099] In some optional embodiments of this embodiment, the extraction module is further configured to perform sentence processing on the text to be processed, and determine the sentence structure information of the text to be processed based on at least one of the following attribute information of the sentence: keyword relevance, number of named entities, sentence length, number of summary keywords, and similarity with key sentences.

[0100] In some optional methods of this embodiment, the semantic information of the text to be processed is obtained by: counting the word frequencies of the words in the text to be processed and sorting them from large to small based on the word frequencies; determining the semantic information of the text to be processed based on a preset number of words that are ranked at the top.

[0101] It should be noted that the collection, collection, updating, analysis, processing, use, transmission, storage and other aspects of user personal information involved in the technical solution of this disclosure are in compliance with the provisions of relevant laws and regulations, are used for legitimate purposes, and do not violate public order and good morals. Necessary measures are taken for user personal information to prevent illegal access to user personal information data and maintain the security of user personal information, network security and national security.

[0102] According to an embodiment of the present application, the present application also provides an electronic device, a readable storage medium and a computer program product.

[0103] like Figure 6 , which is a block diagram of an electronic device according to a summary information generating method according to an embodiment of the present application.

[0104] 600 is a block diagram of an electronic device according to a summary information generation method of an embodiment of the present application. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workbenches, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present application described and / or required herein.

[0105] like Figure 6 As shown, the electronic device includes: one or more processors 601, a memory 602, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. The various components are interconnected using different buses and can be installed on a common mainboard or installed in other ways as needed. The processor can process instructions executed in the electronic device, including instructions stored in or on the memory to display the graphical information of the GUI on an external input / output device (such as a display device coupled to the interface). In other embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories and multiple memories. Similarly, multiple electronic devices can be connected, and each device provides some necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). Figure 6 A processor 601 is taken as an example.

[0106] The memory 602 is a non-transitory computer-readable storage medium provided in the present application. The memory stores instructions executable by at least one processor to enable the at least one processor to perform the summary information generation method provided in the present application. The non-transitory computer-readable storage medium of the present application stores computer instructions, which are used to enable a computer to perform the summary information generation method provided in the present application.

[0107] The memory 602 is a non-transient computer-readable storage medium that can be used to store non-transient software programs, non-transient computer executable programs and modules, such as program instructions / modules corresponding to the summary information generation method in the embodiment of the present application (for example, the attached Figure 5 The processor 601 executes various functional applications and data processing of the server by running the non-transient software programs, instructions and modules stored in the memory 602, that is, the summary information generation method in the above method embodiment is implemented.

[0108] The memory 602 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created by the use of the summary information generating electronic device, etc. In addition, the memory 602 may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory 602 may optionally include a memory remotely arranged relative to the processor 601, and these remote memories may be connected to the summary information generating electronic device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0109] The electronic device of the summary information generation method may further include: an input device 603 and an output device 604. The processor 601, the memory 602, the input device 603 and the output device 604 may be connected via a bus or other means. Figure 6 The example of connecting through bus is taken in the following.

[0110] The input device 603 can receive input digital or character information, such as a touch screen, a keypad, a mouse, a track pad, a touch pad, an indicator bar, one or more mouse buttons, a track ball, a joystick, and other input devices. The output device 604 may include a display device, an auxiliary lighting device (e.g., an LED), and a tactile feedback device (e.g., a vibration motor). The display device may include, but is not limited to, a liquid crystal display (LCD), a light emitting diode (LED) display, and a plasma display. In some embodiments, the display device may be a touch screen.

[0111] Various implementations of the systems and techniques described herein can be realized in digital electronic circuit systems, integrated circuit systems, dedicated ASICs (application specific integrated circuits), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0112] These computer programs (also referred to as programs, software, software applications, or code) include machine instructions for programmable processors and can be implemented using high-level procedural and / or object-oriented programming languages, and / or assembly / machine languages. As used herein, the terms "machine-readable medium" and "computer-readable medium" refer to any computer program product, device, and / or means (e.g., disk, optical disk, memory, programmable logic device (PLD)) for providing machine instructions and / or data to a programmable processor, including a machine-readable medium that receives machine instructions as a machine-readable signal. The term "machine-readable signal" refers to any signal for providing machine instructions and / or data to a programmable processor.

[0113] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0114] The systems and techniques described herein may be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), and the Internet.

[0115] A computer system may include clients and servers. Clients and servers are generally remote from each other and usually interact through a communication network. The relationship of client and server is generated by computer programs running on respective computers and having a client-server relationship to each other.

[0116] According to the technical solution of the embodiment of the present application, the grammatical rationality and semantic integrity of the generated summary are effectively improved.

[0117] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps recorded in this application can be executed in parallel, sequentially or in different orders, as long as the expected results of the technical solution disclosed in this application can be achieved, and this document is not limited here.

[0118] The above specific implementations do not constitute a limitation on the protection scope of this application. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of this application should be included in the protection scope of this application.

Claims

1. A method for generating summary information, the method comprising: Extract sentence structure information of the text to be processed; Based on the sentence structure information and the semantic information of the text to be processed, summary information of the text to be processed is generated.

2. The method according to claim 1, wherein: The step of generating summary information of the text to be processed based on the sentence structure information and the semantic information of the text to be processed includes: The sentence structure information and the semantic information of the text to be processed are input into a generation network to generate summary information of the text to be processed. The generation network is used to characterize the correspondence between the sentence structure information and the semantic information and the summary information.

3. The method according to claim 2, wherein: The generation network includes a first encoding network, a second encoding network and a decoding network, and the sentence structure information and the semantic information of the text to be processed are input into the generation network to generate summary information of the text to be processed, including: Inputting the semantic information into a first encoding network to generate semantic encoding information; Inputting the semantic coding information and the sentence structure information into a second coding network to generate fused coding information, wherein the second coding network is used to encode the sentence structure information, generate structural coding information, and fuse the structural coding information with the semantic coding information; The fused encoding information is input into a decoding network to generate summary information of the text to be processed.

4. The method according to claim 3, wherein: The first coding network includes a first sub-network, a second sub-network and a fusion network, and the inputting the semantic information into the first coding network to generate semantic coding information includes: Inputting the semantic information into the first sub-network to generate forward semantic encoding information; Inputting the semantic information into the second sub-network to generate reverse semantic encoding information; The forward semantic coding information and the reverse semantic coding information are input into a fusion network to generate semantic coding information.

5. The method according to claim 3, wherein: The generation network is a pointer generation network integrated with a convergence mechanism.

6. The method according to any one of claims 1 to 5, wherein: The extracting sentence structure information of the text to be processed includes: The text to be processed is processed into sentences, and the sentence structure information of the text to be processed is determined based on at least one of the following attribute information of the sentence: keyword relevance, number of named entities, sentence length, number of summary keywords, and similarity with key sentences, wherein: The keyword relevance is determined by: using a graph-based sorting algorithm to determine the keywords of the text to be processed, and determining the keyword relevance based on the number of keywords contained in the current sentence; The sentence length is determined by: performing word segmentation processing on the current sentence, and determining the length of the statistically segmented words as the sentence length; The similarity of the key sentences is determined in the following manner: using a graph-based sorting algorithm to determine the key sentences in the text to be processed; and determining the similarity of the key sentences based on the similarity between the current sentence and the key sentences.

7. The method according to any one of claims 1 to 5, wherein: The semantic information of the text to be processed is obtained in the following way: Count the frequency of words in the text to be processed and sort them from large to small based on the frequency; Based on a preset number of top ranked words, semantic information of the text to be processed is determined.

8. A summary information generating device, comprising: An extraction module, configured to extract sentence structure information of the text to be processed; The generating module is configured to generate summary information of the text to be processed based on the sentence structure information and the semantic information of the text to be processed.

9. An electronic device, characterized in that: include: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions executable by the at least one processor so that the at least one processor can perform the method according to any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium storing computer instructions, characterized in that: The computer instructions are used to cause the computer to execute the method according to any one of claims 1 to 7.

11. A computer program product, comprising a computer program, which, when executed by a processor, implements the method according to any one of claims 1 to 7.

Images