Hotspot information detection method and device, equipment, storage medium and program product

Abstract

The invention relates to the technical field of natural language processing, and discloses a hotspot information detection method and device, equipment, a storage medium and a program product. Analyzing the query content to obtain a query word vector of the query content; performing similarity matching on the query word vector and a hot word vector in a dynamic hot word library, wherein the hot word vector is a vector representation of a hot word in the dynamic hot word library; and determining a hot word information detection result of the query content based on a similarity matching result. Therefore, accurate matching of the query content and the hot words in the dynamic hot word library is realized through similarity matching, and the hot word labels are intelligently divided, so that the accuracy and correlation of response of the user to the query content in the hot spot field are effectively improved, and the user experience is remarkably improved.

Description

Technical Field

[0001] The present application relates to the field of natural language processing technology, and specifically to a hotspot information detection method, device, equipment, storage medium and program product. Background Art

[0002] With the explosion of information and the rapid development of social media, hot topics are updated very frequently, and users' focus is constantly changing. Traditional question-answering systems cannot respond promptly and accurately to real-time hot topics, resulting in delayed or irrelevant information obtained by users. Therefore, a data processing technology that can dynamically adapt to real-time hot topic changes is needed to improve the ability to identify and respond to the latest hot content.

[0003] In this regard, in the related technologies, hotspot detection based on keyword matching, hotspot identification models based on historical data training, and fixed label distribution mechanisms are mainly used to realize the recognition and response of real-time hot content. However, hotspot detection based on keyword matching needs to detect real-time hot content through preset keywords or phrases. It is easy to miss or misjudge when facing emerging vocabulary or semantic changes, and it is difficult to adapt to the dynamically updated hot environment. The hotspot identification model based on historical data training uses a large amount of historical data to train the model to predict and identify content that may become hotspots. It has a good processing effect on hotspots in known fields, but it responds slowly to emergencies or new topics and cannot reflect the latest hot trends in real time. The fixed label distribution mechanism allocates labels based on predefined categories or topics, lacks dynamic perception of real-time hotspots, and is difficult to flexibly adjust the applicability of labels according to semantic similarity, especially in scenarios with high real-time requirements. Therefore, how to realize timely and accurate recognition and response to hot content has become a technical problem that needs to be solved urgently. Summary of the invention

[0004] In view of this, the present application provides a hotspot information detection method, device, equipment, storage medium and program product to solve the technical problem of how to achieve timely and accurate identification and response to hotspot content, which has become an urgent problem to be solved.

[0005] In a first aspect, the present application provides a hotspot information detection method, the method comprising:

[0006] Receive user query content;

[0007] Parse the query content to obtain the query word vector of the query content;

[0008] Perform similarity matching between the query word vector and the hot word vector in the dynamic hot word library. The hot word vector is the vector representation of the hot word in the dynamic hot word library.

[0009] Based on the similarity matching results, the hot word information detection results of the query content are determined.

[0010] The hot information detection method of this application parses the user's query content, directly performs similarity matching between the query word vector obtained by the analysis and the hot word vector in the dynamic hot word library, and determines the hot word label that matches the query content based on the similarity matching result. Thus, through similarity matching, the query content is accurately matched with the hot words in the dynamic hot word library, and the hot word labels are intelligently distributed, effectively improving the accuracy and relevance of the user's response to the query content in the hot field, and significantly improving the user experience.

[0011] In an optional implementation, parsing the query content includes:

[0012] Perform word segmentation, part-of-speech tagging and semantic analysis on the query content to obtain semantic analysis results;

[0013] Vectorize the semantic parsing results to generate query word vectors.

[0014] The hot information detection method of the present application obtains semantic analysis results by performing word segmentation, part-of-speech tagging and semantic analysis on the query content, and vectorizes the semantic analysis results to generate query word vectors, thereby maximizing the consistency between the user's query content and hot words, thereby effectively ensuring the accuracy of the similarity matching results.

[0015] In an optional implementation, similarity matching is performed between the query word vector and the hot word vector in the dynamic hot word library, including:

[0016] Filter the heat value of the hot word vector in the dynamic hot word library whose similarity with the query word vector is greater than the set similarity threshold.

[0017] In an optional implementation, based on the similarity matching result, determining the hot word information detection result of the query content includes:

[0018] Get the heat value of the hot word vector whose similarity with the query word vector is greater than the set similarity threshold;

[0019] Get the user's historical query data;

[0020] According to the popularity value and historical query data, based on the set weight rules, the hot word information detection results of the query content are determined.

[0021] The hot information detection method of this application combines the user's historical query data and similarity matching results, and based on the pre-set weight rules, assigns hot word tags suitable for the user who issued the query content to the corresponding query content. It effectively improves the accuracy of hot information detection, thereby significantly improving the accuracy and pertinence of responding to the user's query content, and significantly improving the user experience.

[0022] In an optional implementation, after determining the hot word information detection result of the query content, the method further includes:

[0023] Based on the update of the dynamic hot word library, the hot word tags matching the query content are updated.

[0024] The hot word information detection method of the present application updates the dynamic hot word library in real time, and when the dynamic hot word library is updated, the hot word tags matching the query content are updated in time, further improving the real-time performance of hot word information detection for the user's query content.

[0025] In an optional implementation, after determining the hot word information detection result of the query content, the method further includes:

[0026] Detect user behavior data;

[0027] Based on the behavioral data, update the hot word tags that match the query content.

[0028] The hot word information detection method of the present application detects the user's behavior data and updates the hot word labels based on the behavior data. Thus, the hot word labels obtained by the hot word information detection of the query content are adjusted in a timely manner according to the feedback of the user's behavior data, so as to record accurate and effective user historical query data, provide an effective reference for subsequent hot word information detection, and effectively improve the accuracy of subsequent hot word information detection.

[0029] In an optional implementation, before receiving the user's query content, the method further includes:

[0030] Monitor the hotspot information of the set data source;

[0031] Preprocess hotspot information;

[0032] Vectorize the preprocessed hotspot information to generate high-dimensional word vectors;

[0033] Aggregate high-dimensional word vectors to obtain hot word vectors in the dynamic hot word library.

[0034] In an optional implementation, after aggregating the high-dimensional word vectors, the method further includes:

[0035] Use sliding window or incremental update strategy to dynamically update the dynamic hot word library.

[0036] The hot information detection method of this application dynamically monitors and calculates the hot word vectors of real-time hot information in the data source through an algorithm, ensuring that the hot information of the latest dynamic hot word library can be used when processing user query content, thereby effectively improving the ability to capture current events and trends in the query content.

[0037] In a second aspect, the present application provides a hotspot information detection device, the device comprising:

[0038] A receiving module, used for receiving the query content of the user;

[0039] A parsing module is used to parse the query content and obtain the query word vector of the query content;

[0040] The matching module is used to perform similarity matching between the query word vector and the hot word vector in the dynamic hot word library;

[0041] The determination module is used to determine the hot word information detection result of the query content based on the similarity matching result.

[0042] In a third aspect, the present application provides a computer device, comprising: a memory and a processor, the memory and the processor are communicatively connected to each other, computer instructions are stored in the memory, and the processor executes the hotspot information detection method of the above-mentioned first aspect or any corresponding embodiment thereof by executing the computer instructions.

[0043] In a fourth aspect, the present application provides a computer-readable storage medium having computer instructions stored thereon, the computer instructions being used to enable a computer to execute the hotspot information detection method of the first aspect or any corresponding embodiment thereof.

[0044] In a fifth aspect, the present application provides a computer program product, including computer instructions, which are used to enable a computer to execute the hotspot information detection method of the above-mentioned first aspect or any corresponding embodiment thereof.

[0045] The hot information detection method, device, equipment, storage medium and program product of the present application parse the query content of the user, directly perform similarity matching between the query word vector obtained by the analysis and the hot word vector in the dynamic hot word library, and determine the hot word label that matches the query content based on the similarity matching result. Thus, through similarity matching, the query content is accurately matched with the hot words in the dynamic hot word library, and the hot word labels are intelligently distributed, which effectively improves the accuracy and relevance of the user's response to the query content in the hot field, and significantly improves the user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to more clearly illustrate the specific implementation methods of the present application or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0047] Figure 1 is a flow chart of a hotspot information detection method according to an embodiment of the present application;

[0048] Figure 2 is a flow chart of another hotspot information detection method according to an embodiment of the present application;

[0049] Figure 3 is a flowchart of a specific application example of the hotspot information detection method according to an embodiment of the present application;

[0050] Figure 4 is a structural block diagram of a hotspot information detection device according to an embodiment of the present application;

[0051] Figure 5 It is a schematic diagram of the hardware structure of the computer device of an embodiment of the present application. DETAILED DESCRIPTION

[0052] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without making creative work are within the scope of protection of this application.

[0053] How to achieve timely and accurate identification and response to hot content has become a technical problem that needs to be solved urgently. The present application provides a hot information detection method, device, equipment, storage medium and program product, which are applied to servers, etc. The server can configure large file systems, etc. The present application provides a hot information detection method based on a dynamic label distribution algorithm of multi-source data fusion and semantic similarity calculation, which effectively improves the response accuracy and flexibility of the question-answering system in a rapidly changing environment by updating hot word vectors in real time, intelligently matching user queries, and adaptively distributing labels.

[0054] According to an embodiment of the present application, an embodiment of a hotspot information detection method is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0055] In an embodiment of the present application, a hotspot information detection method is provided, which can be used for the above-mentioned server, etc. Figure 1 is a flow chart of a hotspot information detection method according to an embodiment of the present application. Figure 1 As shown, the process includes the following steps:

[0056] Step S101, receiving the user's query content.

[0057] In some optional implementations, the user's query content may be a word or a sentence, or may be other suitable forms of content, usually referred to as query.

[0058] Step S102, parse the query content to obtain a query word vector of the query content.

[0059] In some optional implementations, the query content input by the user may be segmented, POS-tagged, and semantically analyzed, and the query word vector may be generated from the analyzed content.

[0060] In order to facilitate the operation of performing similarity matching between the query word vector and the hot word vector in the dynamic hot word library in the following step S103, the operation of generating the query word vector here can generate a vectorization model used by the hot word vector in the dynamic hot word library, such as the BERT (Bidirectional Encoder Representations from Transformers) model.

[0061] The BERT model is a Transformer-based bidirectional semantic encoding model that uses the encoder part of the Transformer instead of the decoder part. BERT learns rich language representations by pre-training on large-scale unlabeled corpora, which can then be fine-tuned in various downstream tasks to adapt to specific needs. The BERT model mainly consists of three modules: the Embedding (a technology that maps high-dimensional data to low-dimensional space) module, the Transformer (deep learning based on attention mechanism) module, and the pre-training and fine-tuning module. The BERT model has wide applicability and strong performance in natural language processing in application scenarios such as reading comprehension, text classification, sentiment analysis, and named entity recognition.

[0062] Step S103, performing similarity matching between the query word vector and the hot word vector in the dynamic hot word library, where the hot word vector is a vector representation of the hot word in the dynamic hot word library.

[0063] In some optional implementations, algorithms such as cosine similarity, Euclidean distance, or KL (Kullback-Leibler, reference distribution) divergence can be used to calculate the similarity between the query word vector and the hot word vector in the dynamic hot word library. Based on the similarity calculation result, the hot word vector in the dynamic hot word library whose similarity with the query word vector is greater than the set similarity threshold can be screened out.

[0064] Step S104: determining a hot word information detection result of the query content based on the similarity matching result.

[0065] If no hot word vectors whose similarity with the query word vector in the dynamic hot word library is greater than the set similarity threshold are screened out in step S103, that is, if the number of hot word vectors whose similarity with the query word vector in the dynamic hot word library is greater than the set similarity threshold is zero, feedback information is returned to indicate that the user's query content has no matching hot word information. A dialogue system or the like using the hot word information detection method of the embodiment of the present application can respond to the user's query content with a default result or perform other set operations.

[0066] If the hot word vectors whose similarity with the query word vector is greater than the set similarity threshold are screened out in the dynamic hot word library in step S103, that is, if the number of hot word vectors whose similarity with the query word vector is greater than the set similarity threshold in the dynamic hot word library is greater than zero, it is determined that the user's query content is related to the corresponding hot information, and the corresponding hot information is the hot word corresponding to the hot word vector whose similarity with the query word vector is greater than the set similarity threshold. Here, hot word labels can be distributed for the user's query content related to the corresponding hot information.

[0067] The hot information detection method of this application parses the user's query content, directly performs similarity matching between the query word vector obtained by the analysis and the hot word vector in the dynamic hot word library, and determines the hot word label that matches the query content based on the similarity matching result. Thus, through similarity matching, the query content is accurately matched with the hot words in the dynamic hot word library, and the hot word labels are intelligently distributed, effectively improving the accuracy and relevance of the user's response to the query content in the hot field, and significantly improving the user experience.

[0068] In an embodiment of the present application, a hotspot information detection method is provided, which can be used for the above-mentioned server, etc. Figure 2 is a flow chart of a hotspot information detection method according to an embodiment of the present application. Figure 2 As shown, the process includes the following steps:

[0069] Step S201, monitoring hotspot information of a set data source.

[0070] In some optional implementations, the set data source may include set social media, news websites, etc. The specific data sources used as the set data sources may be set and adjusted according to actual needs. Here, a crawler or the like may be used to collect hot information in real time from set social media, news websites, and user behavior data.

[0071] Step S202, pre-processing the hotspot information.

[0072] In order to process the hot information more accurately, the hot information is first pre-processed by filtering. Specifically, irrelevant content can be filtered, and word segmentation, duplication removal and stop word processing can be performed.

[0073] Step S203, vectorize the preprocessed hotspot information to generate a high-dimensional word vector.

[0074] Here, models such as Word2Vec and BERT can be used to vectorize the processed data to generate high-dimensional word vector representations.

[0075] Word2Vec is a model for generating word vectors. It can learn semantic knowledge from a large amount of text corpus through unsupervised learning, converting each word in the vocabulary into a high-dimensional vector that can represent the semantic relationship between words. The BERT model is a bidirectional semantic encoding model based on Transformer, using the encoder part of Transformer.

[0076] Step S204, aggregate the high-dimensional word vectors to obtain hot word vectors in the dynamic hot word library.

[0077] Here, the high-dimensional word vectors generated in step S203 can be subjected to aggregation operations such as weighted averaging and TF-IDF weighting, so as to perform deduplication and other processing on multiple high-dimensional word vectors to obtain a comprehensive vector representation of hot topics. Each comprehensive vector can be used as a hot word vector, and multiple comprehensive vectors can be used as a dynamic hot word library.

[0078] Step S205, dynamically update the dynamic hot word library using a sliding window or incremental update strategy.

[0079] Here, the hot word vectors in the dynamic hot word library can be updated regularly at set intervals to remove the hot word vectors and corresponding hot information that are no longer timely. Specifically, a sliding window or incremental update strategy can be used to maintain the real-time nature of the vector, and the hot word vectors can be stored in a cache or database so that the dynamic hot word library can be called when receiving the user's query content.

[0080] The sliding window strategy is an incremental learning technique that divides the training set into overlapping windows. As new data becomes available, the sliding window moves forward, discarding the oldest data and adding the latest data. This ensures that the training set always contains the latest information while limiting the amount of data the model needs to be retrained on.

[0081] The incremental update strategy is a machine learning paradigm that allows a model to incrementally learn new data without retraining the entire dataset and can be used to dynamically update the training set to include newly collected or labeled data.

[0082] This application dynamically calculates the hot word vectors of real-time hot information to ensure that the latest hot information is used when processing the user's query content (query), which significantly improves the ability to capture current events and trends.

[0083] In the implementation process of the hotspot information detection method of the embodiment of the present application, the above steps S201 to S205 can be used as a processing module. For example, the above steps S201 to S205 are implemented by the following code:

[0084] #Module 1: Real-time hot word vector calculation

[0085] class HotspotVectorCalculator:

[0086] Initialize with hotspot data

[0087] # Initialize the model using hotspot data

[0088] Preprocess data(tokenization,filtering)

[0089] #Preprocessing data: word segmentation and filtering irrelevant content

[0090] Train Word2Vec model on processed data

[0091] #Train the Word2Vec model on the processed data

[0092] For each topic:

[0093] Generate vector by aggregating word vectors

[0094] #Generate word vectors for each topic, and get the topic vector by aggregating word vectors

[0095] Return hotspot vectors

[0096] #Return the word vectors of all hot spots

[0097] Step S206: receiving the user's query content.

[0098] For details, please see Figure 1 Step S101 of the illustrated embodiment will not be described in detail here.

[0099] Step S207, parse the query content to obtain a query word vector of the query content.

[0100] In an optional implementation, step S207 may include:

[0101] Step S2071, perform word segmentation, part-of-speech tagging and semantic analysis on the query content to obtain a semantic analysis result.

[0102] Step S2072, vectorize the semantic analysis result to generate a query word vector.

[0103] The hot information detection method of the present application obtains semantic analysis results by performing word segmentation, part-of-speech tagging and semantic analysis on the query content, and vectorizes the semantic analysis results to generate query word vectors, thereby maximizing the consistency between the user's query content and hot words, thereby effectively ensuring the accuracy of the similarity matching results.

[0104] Step S208, performing similarity matching between the query word vector and the hot word vector in the dynamic hot word library, where the hot word vector is a vector representation of the hot word in the dynamic hot word library.

[0105] In an optional implementation, the heat value of the hot word vector whose similarity with the query word vector is greater than a set similarity threshold in the dynamic hot word library can be screened, thereby implementing the above step S208.

[0106] In the implementation process of the hotspot information detection method of the embodiment of the present application, the above steps S206 to S208 can be used as a processing module. For example, the above steps S206 to S208 are implemented by the following code:

[0107] #Module 2: User query semantic analysis and similarity calculation

[0108] class QueryMatcher:

[0109] Initialize with hotspot vectors

[0110] # Initialize the query matcher using the hotspot vector

[0111] For each user query:

[0112] Tokenize query using BERT tokenizer

[0113] #Use BERT tokenizer to tokenize user queries

[0114] Generate query vector using BERT model

[0115] #Use the BERT model to generate vector representations of user queries

[0116] Calculate similarity between query vector and each hotspot vector

[0117] #Calculate the similarity between the user query vector and each hotspot vector

[0118] If similarity>threshold:

[0119] Return matched topics

[0120] #If the similarity exceeds the set threshold, the matching hot topics are returned

[0121] Step S209: determining the hot word information detection result of the query content based on the similarity matching result.

[0122] In an optional implementation, step S209 may include:

[0123] Step S2091, obtaining the heat value of the hot word vector whose similarity with the query word vector is greater than the set similarity threshold.

[0124] Here, the heat value of the hot word vector can be obtained when the hot word vector in the dynamic hot word library is obtained, and the heat value of each hot word vector is determined at the same time, recorded and stored in the dynamic hot word library. For example, the heat value can be a value of 0 to 10. 0 represents the lowest heat and 10 represents the highest heat.

[0125] Step S2092, obtaining the user's historical query data.

[0126] The user's historical query data here can be set according to actual needs. For example, the hot information detection method of this application is applied to music search, navigation, product browsing, etc. The historical query data may include songs that the user has listened to, locations that the user has searched or navigated, and products that the user has searched or purchased.

[0127] Step S2093, according to the popularity value and historical query data, based on the set weight rules, determine the hot word information detection result of the query content.

[0128] Here, the user's historical preferences can be determined based on the user's historical query data and quantified. For example, the hot words that have appeared in the user's historical query data and the hot words that have not appeared, for the user, the corresponding hot word vectors in the corresponding dynamic hot word library are respectively assigned values ​​for representing the user's historical preferences.

[0129] Based on the similarity matching result of the query word vector and the hot word vector in the dynamic hot word library, the hot word label related to the user query is assigned in combination with the preset weight rules, such as the heat value, the user's historical preference, etc. For example: when the value is assigned to 1, the historical query data representing the user includes the query content with the hot word label, and when the value is assigned to 0, the historical query data representing the user does not include the query content with the hot word label.

[0130] For example, among the multiple hot word vectors whose similarity with the query word vector is greater than the set similarity threshold, there is a hot word vector that matches the query word vector of the user's query content with an 80% match, the user preference of the hot word vector is 1, the heat value of the hot word is 9, and the set weight rule is that the weight of the hot word vector's matching degree is 80%, and the weight of the heat value is 20%. The processing method of the user's historical preference is that if the weighted calculation result of the matching degree and the heat value is greater than the first threshold, the user's historical preference is not considered, and the hot information corresponding to the hot word vector is directly used as the user's hot word label. If the weighted calculation result of the matching degree and the heat value is less than the second threshold and greater than the third threshold, then when the user's historical preference is 1, the hot information corresponding to the hot word vector is used as the user's hot word label, and when the user's historical preference is 0, the hot information corresponding to the hot word vector is no longer used as the user's hot word label. If the weighted calculation result of the matching degree and the heat value is less than the third threshold, regardless of whether the user's historical preference is 1 or 0, the hot information corresponding to the hot word vector is no longer used as the user's hot word label.

[0131] The hot information detection method of this application combines the user's historical query data and similarity matching results, and based on the pre-set weight rules, assigns hot word tags suitable for the user who issued the query content to the corresponding query content. It effectively improves the accuracy of hot information detection, thereby significantly improving the accuracy and pertinence of responding to the user's query content, and significantly improving the user experience.

[0132] Step S210, detecting user behavior data.

[0133] Here, the user's behavior data mainly refers to the distribution of hot word tags for the user's query content and the way the user handles the response content after responding to the user's query content. For example, the user clicks on the corresponding pushed content and stays to view it for more than the set time threshold, the user clicks the button used to represent cancellation or return, etc. If the user clicks on the corresponding pushed content and stays to view it for more than the set time threshold, it can be determined that there may be deviations in the hot word tags. If the user clicks the button used to represent cancellation or return, it can be determined that the user's hot word tags are relatively accurate. Here you can also obtain information such as the specific time when the user viewed the corresponding content.

[0134] Step S211, updating hot word tags according to the behavior data.

[0135] Based on the user's behavior data detected in step S210, it can be determined whether there is a deviation in the hot word tag, and the probability of the deviation. Specifically, the judgment conditions can be set according to actual needs. When the probability of the hot word tag having a deviation is greater than the set probability, the hot word tag can be deleted. When the probability of the hot word tag having a deviation is not greater than the set probability, the hot word tag can be retained. In this way, the hot spot changes are monitored in real time, and the distributed tags are adjusted and optimized.

[0136] Furthermore, hot word tags and log records can be output. Specifically, the final hot word tags can be generated and output, and the distribution process and results can be recorded at the same time, which is convenient for subsequent analysis and optimization, and realizes continuous improvement of the hot word information detection method of this application.

[0137] In the implementation process of the hotspot information detection method of the embodiment of the present application, the above steps S209 to S211 can be used as a processing module. For example, the above steps S209 to S211 are implemented by the following code:

[0138] #Module 3: Dynamic label distribution and management

[0139] class TagDistributor:

[0140] Initialize with matched topics

[0141] #Initialize the tag distributor using matching hot topics

[0142] Initialize tag database

[0143] # Initialize the tag database

[0144] For each matched topic:

[0145] Retrieve and aggregate relevant tags from tag database

[0146] #For each matching topic, retrieve and summarize the relevant tags from the tag database

[0147] Return distributed tags

[0148] #Return the distribution label

[0149] #Main process

[0150] Load hotspot data

[0151] #Load hotspot data

[0152] Process hotspot data with HotspotVectorCalculator

[0153] #Use HotspotVectorCalculator to process hotspot data and generate word vectors

[0154] Generate query vector using QueryMatcher

[0155] #Use QueryMatcher to generate user query vectors

[0156] Match query with relevant hotspots

[0157] #Match user queries with relevant hotspots

[0158] Distribute tags using TagDistributor

[0159] #Use TagDistributor to distribute tags

[0160] Output final tags

[0161] # Output the final label

[0162] In an embodiment of the present application, a hotspot information detection method is provided, which can be used for the above-mentioned server, etc. Figure 3 is a flowchart of a specific application example of the hotspot information detection method according to an embodiment of the present application, such as Figure 3 As shown, the process includes the following steps:

[0163] Step S301, monitoring hotspot information of a set data source.

[0164] Step S302, pre-processing the hotspot information.

[0165] Step S303: training a vectorized model based on the preprocessed data.

[0166] Here, the Word2Vec, BERT and other model architectures of step S203 can be used.

[0167] Step S304: Generate hot word vectors of the dynamic hot word library based on the vectorization model.

[0168] The above steps S301 to S304 can be executed repeatedly in a loop, wherein the operation of training the vectorized model based on the preprocessed data in step S303 can be the initial training of the model, or can be the self-learning of the vectorized model based on the update of the hot information. Therefore, the hot word vectors of the dynamic hot word library are dynamically updated with the acquired hot information.

[0169] Step S305: receiving the user's query content.

[0170] Step S306, calculating the query word vector of the user's query content, and calculating the similarity between the query word vector and the hot word vector.

[0171] Step S307, performing similarity judgment based on the similarity calculation results between the query word vector and the hot word vector.

[0172] When the similarity judgment result shows that there is a hot word vector that matches the query word vector, step S308 is executed. When the similarity judgment result shows that there is no hot word vector that matches the query word vector, step S310 is executed.

[0173] The above steps S306 to S307 may refer to the operations such as "matching the query word vector with the hot word vector in the dynamic hot word library in similarity" in the above steps S103, S104 and S208.

[0174] Step S308: determining hot word tags of the query content based on the similarity judgment result.

[0175] Step S309: Output tags based on user historical query data.

[0176] The tags here are hot word tags that match the user's query content and are determined and dynamically adjusted based on the user's historical query data and similarity matching results.

[0177] Step S310: output information for indicating no hit.

[0178] Specifically, no hit means that the similarity judgment result shows that there is no hot word vector matching the query word vector.

[0179] Figure 3 For other detailed implementation details of the illustrated embodiment, please refer to the above Figure 1 and Figure 2 The detailed implementation details of the illustrated embodiment will not be repeated here.

[0180] The hot information detection method of this application dynamically monitors and calculates the hot word vectors of real-time hot information in the data source through an algorithm, ensuring that the hot information of the latest dynamic hot word library can be used when processing user query content, thereby effectively improving the ability to capture current events and trends in the query content.

[0181] In the embodiments of the present application, a hotspot information detection device is also provided, which is used to implement the above embodiments and preferred implementation modes, and the descriptions that have been made will not be repeated. As used below, the term "module" can implement a combination of software and / or hardware for a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.

[0182] The present application embodiment provides a hot spot information detection device, such as Figure 4 As shown, including:

[0183] Receiving module 401, used to receive the query content of the user;

[0184] A parsing module 402 is used to parse the query content to obtain a query word vector of the query content;

[0185] A matching module 403 is used to perform similarity matching between the query word vector and the hot word vector in the dynamic hot word library;

[0186] The determination module 404 is used to determine the hot word information detection result of the query content based on the similarity matching result.

[0187] In an optional implementation, the parsing module 402 includes:

[0188] Semantic unit, used to perform word segmentation, part-of-speech tagging and semantic analysis on the query content to obtain semantic analysis results;

[0189] The vector unit is used to vectorize the semantic parsing results and generate query word vectors.

[0190] In an optional implementation, the matching module 403 includes:

[0191] The screening unit is used to screen the heat value of the hot word vector whose similarity with the query word vector in the dynamic hot word library is greater than a set similarity threshold.

[0192] In an optional implementation, the determination module 404 includes:

[0193] Get the heat value of the hot word vector whose similarity with the query word vector is greater than the set similarity threshold;

[0194] Get the user's historical query data;

[0195] According to the popularity value and historical query data, based on the set weight rules, the hot word information detection results of the query content are determined.

[0196] In an optional embodiment, the device further comprises:

[0197] The first updating module is used to update the hot word tags matching the query content based on the update of the dynamic hot word library after determining the hot word information detection result of the query content.

[0198] In an optional embodiment, the device further comprises:

[0199] A user detection module is used to detect user behavior data after determining the hot word information detection result of the query content;

[0200] The second updating module is used to update the hot word labels according to the behavior data.

[0201] In an optional embodiment, the device further comprises:

[0202] A monitoring module is used to monitor hotspot information of a set data source before receiving the query content of the user;

[0203] A preprocessing module, used for preprocessing hotspot information;

[0204] The vector module is used to vectorize the preprocessed hotspot information and generate high-dimensional word vectors;

[0205] The aggregation module is used to aggregate high-dimensional word vectors to obtain hot word vectors in the dynamic hot word library.

[0206] In an optional embodiment, the device further comprises:

[0207] The vocabulary update module is used to dynamically update the dynamic hot vocabulary by adopting a sliding window or incremental update strategy after aggregating the high-dimensional word vectors.

[0208] The further functional description of each of the above modules and units is the same as that of the above corresponding embodiments and will not be repeated here.

[0209] The hotspot information detection device in the embodiment of the present application is presented in the form of a functional unit, where the unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that executes one or more software or fixed programs, and / or other devices that can provide the above functions.

[0210] The present application also provides a computer device having the above Figure 4 The hotspot information detection device shown.

[0211] See also Figure 5 , Figure 5 is a schematic diagram of the structure of a computer device provided by an optional embodiment of the present application, such as Figure 5 As shown, the computer device includes: one or more processors 10, a memory 20, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. Various components are connected to each other using different buses for communication, and can be installed on a common mainboard or installed in other ways as needed. The processor can process the instructions executed in the computer 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 some optional embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories and multiple memories. Similarly, multiple computer 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 5 A processor 10 is taken as an example.

[0212] The processor 10 may be a central processing unit, a network processor or a combination thereof. The processor 10 may further include a hardware chip. The hardware chip may be a dedicated integrated circuit, a programmable logic device or a combination thereof. The programmable logic device may be a complex programmable logic device, a field programmable gate array, a general purpose array logic or any combination thereof.

[0213] The memory 20 stores instructions executable by at least one processor 10, so that at least one processor 10 executes the method shown in the above embodiment.

[0214] The memory 20 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 according to the use of the computer device, etc. In addition, the memory 20 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 optional embodiments, the memory 20 may optionally include a memory remotely arranged relative to the processor 10, and these remote memories may be connected to the computer 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.

[0215] The memory 20 may include a volatile memory, such as a random access memory; the memory may also include a non-volatile memory, such as a flash memory, a hard disk or a solid state drive; the memory 20 may also include a combination of the above types of memory.

[0216] The computer device further comprises a communication interface 30 for the computer device to communicate with other devices or a communication network.

[0217] The embodiment of the present application also provides a computer-readable storage medium. The above method according to the embodiment of the present application can be implemented in hardware, firmware, or can be implemented as a computer code that can be recorded in a storage medium, or can be implemented as a computer code that is originally stored in a remote storage medium or a non-temporary machine-readable storage medium and will be stored in a local storage medium through a network download, so that the method described herein can be stored in such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only storage memory, a random access memory, a flash memory, a hard disk or a solid-state hard disk, etc.; further, the storage medium can also include a combination of the above types of memories. It can be understood that a computer, a processor, a microprocessor controller, or programmable hardware includes a storage component that can store or receive software or computer code. When the software or computer code is accessed and executed by a computer, a processor, or hardware, the method shown in the above embodiment is implemented.

[0218] Part of the present application may be applied as a computer program product, such as a computer program instruction, which, when executed by a computer, can call or provide the method and / or technical solution according to the present application through the operation of the computer. Those skilled in the art should understand that the existence of computer program instructions in computer-readable media includes but is not limited to source files, executable files, installation package files, etc., and accordingly, the way in which computer program instructions are executed by a computer includes but is not limited to: the computer directly executes the instruction, or the computer compiles the instruction and then executes the corresponding compiled program, or the computer reads and executes the instruction, or the computer reads and installs the instruction and then executes the corresponding installed program. Here, the computer-readable medium can be any available computer-readable storage medium or communication medium accessible to the computer.

[0219] Although the embodiments of the present application are described in conjunction with the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present application, and such modifications and variations are all within the scope defined by the appended claims.

Claims

1. A hotspot information detection method, characterized in that: The method comprises: Receive user query content; Parsing the query content to obtain a query word vector of the query content; Perform similarity matching between the query word vector and the hot word vector in the dynamic hot word library, where the hot word vector is a vector representation of the hot word in the dynamic hot word library; Based on the similarity matching result, a hot word information detection result of the query content is determined.

2. The method according to claim 1, characterized in that The parsing of the query content includes: Performing word segmentation, part-of-speech tagging and semantic analysis on the query content to obtain a semantic analysis result; The semantic parsing result is vectorized to generate the query word vector.

3. The method according to claim 1, characterized in that The similarity matching of the query word vector with the hot word vector in the dynamic hot word library includes: The heat value of the hot word vector whose similarity with the query word vector in the dynamic hot word library is greater than a set similarity threshold is screened.

4. The method according to claim 1, characterized in that The step of determining the hot word information detection result of the query content based on the similarity matching result includes: Obtaining the heat value of the hot word vector whose similarity with the query word vector is greater than a set similarity threshold; Obtaining historical query data of the user; According to the popularity value and the historical query data, based on the set weight rule, the hot word information detection result of the query content is determined.

5. The method according to claim 1, characterized in that After determining the hot word information detection result of the query content, the method further includes: Based on the update of the dynamic hot word library, the hot word tags matching the query content are updated.

6. The method according to claim 5, characterized in that After determining the hot word information detection result of the query content, the method further includes: Detecting behavioral data of the user; According to the behavior data, the hot word tags matching the query content are updated.

7. The method according to any one of claims 1 to 6, characterized in that Before receiving the query content of the user, the method further includes: Monitor the hotspot information of the set data source; Preprocessing the hotspot information; Vectorize the preprocessed hotspot information to generate high-dimensional word vectors; The high-dimensional word vectors are aggregated to obtain hot word vectors in the dynamic hot word library.

8. The method according to claim 7, characterized in that After aggregating the high-dimensional word vectors, the method further includes: The dynamic hot word library is dynamically updated by adopting a sliding window or incremental update strategy.

9. A hotspot information detection device, characterized in that: The device comprises: A receiving module, used for receiving the query content of the user; A parsing module, used to parse the query content to obtain a query word vector of the query content; A matching module, used for performing similarity matching between the query word vector and the hot word vector in the dynamic hot word library, wherein the hot word vector is a vector representation of the hot word in the dynamic hot word library; The determination module is used to determine the hot word information detection result of the query content based on the similarity matching result.

10. A computer device, characterized in that: include: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the hotspot information detection method according to any one of claims 1 to 8 by executing the computer instructions.

11. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a computer to execute the hotspot information detection method according to any one of claims 1 to 8.

12. A computer program product, characterized in that It includes computer instructions, and the computer instructions are used to enable a computer to execute the hot spot information detection method described in any one of claims 1 to 8.

Images