Method, device and equipment for analyzing hot information of an enterprise

Abstract

This application provides a method, apparatus, and equipment for analyzing trending information about enterprises, relating to the field of data processing technology. The method includes: acquiring the target enterprise's configured batch processing rules for trending topics; collecting first trending content related to the target enterprise from multiple platforms based on trending topic search rules; processing the first trending content based on trending content processing rules to obtain second trending content; analyzing the second trending content to obtain trending analysis results; matching the target public relations strategy corresponding to the influence type from a configured table of different influence types and public relations strategies; and generating public relations suggestions for the target enterprise based on the second trending content and the target public relations strategy. This application enables companies to collect and analyze trending content in a targeted manner to determine the type of influence of the trending topics on the corresponding companies, allowing them to take timely public relations measures.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and more specifically, to a method, apparatus, and equipment for analyzing hot information of an enterprise. Background Technology

[0002] In the era of information overload, enterprises face unprecedented challenges in analyzing trending information. With the rapid development of digital technology, multimodal data (including text, images, and videos) on various platforms is growing exponentially, while the speed and impact of online trending events have significantly increased. Against this backdrop, how enterprises can build efficient content processing mechanisms to accurately capture and predict trends in enterprise-related trending events has become a critical issue that urgently needs to be addressed.

[0003] Existing content processing systems suffer from numerous shortcomings. Content library functions are heavily coupled, and the logic for content capture, processing, and display is mixed, leading to difficult system maintenance and low processing efficiency. In terms of content key point extraction, traditional algorithms struggle to comprehensively and accurately extract crucial information. In hot topic analysis, the relevance of a hot topic to a company still requires manual judgment. Furthermore, the capture of hot topics is neither timely nor accurate, making it impossible to effectively predict trends and hindering companies from implementing timely public relations strategies to expand their influence or avoid negative impacts. Summary of the Invention

[0004] The purpose of this application is to provide a method, apparatus, and device for analyzing enterprise hotspot information, in order to solve the technical problems of low content processing efficiency, untimely and inaccurate hotspot capture, and difficulty in processing multimodal content in the prior art. It can realize efficient multi-dimensional analysis and processing of multimodal content, intelligently track hotspots, and significantly improve data processing efficiency.

[0005] In a first aspect, the present invention provides a method for analyzing hot topic information of enterprises, the method comprising:

[0006] Obtain the hotspot batch processing rules configured by the target enterprise; wherein, the hotspot batch processing rules include: hotspot search rules and hotspot content processing rules;

[0007] Based on the aforementioned hotspot search rules, first hotspot content related to the target company is collected from multiple platforms;

[0008] Based on the aforementioned hot topic content processing rules, the first hot topic content is processed to obtain the second hot topic content;

[0009] The second hot topic is analyzed to obtain hot topic analysis results; wherein, the hot topic analysis results include: the type of impact of the second hot topic on the target enterprise;

[0010] Match the target public relations strategy corresponding to the different impact types from the configured table of different impact types and public relations strategies;

[0011] Based on the second hot topic and the target public relations strategy, public relations recommendations are generated for the target company.

[0012] In an optional implementation, the hot search rules include: hot keywords, filtering rules, at least one search platform, and the search strategy of the corresponding search platform.

[0013] In an optional implementation, the step of collecting first trending content related to the target enterprise from multiple platforms based on the trending search rules includes:

[0014] For any search platform, based on the search strategy of the search platform, retrieve initial search content associated with the hot keywords from the search platform;

[0015] Calculate the correlation between the initial search content and the hot keywords; if the correlation is greater than the configured correlation threshold, then use the initial search content as the first search content.

[0016] Sort the top search results on each search platform in descending order of relevance.

[0017] The sorted first search results are filtered using the aforementioned filtering rules to obtain the second search results;

[0018] The second search result is identified as the top trending topic.

[0019] In an optional implementation, processing the first hot topic content based on the hot topic content processing rules to obtain the second hot topic content includes:

[0020] Based on the aforementioned hot topic content processing rules, the first hot topic content is segmented using a pre-trained word graph model to obtain multiple keywords;

[0021] The first hot topic content is input into a pre-trained sentiment analysis model based on BERT fine-tuning to obtain the sentiment tendency of the first hot topic content and the confidence level of the corresponding sentiment tendency.

[0022] The obtained keywords, sentiment tendencies and corresponding sentiment tendency confidence scores, entity recognition results, and the first hot topic content are input into a pre-trained attention-enhanced sequence-based summary generation model to obtain a hot topic summary of the first hot topic content.

[0023] Based on the hot topic summary, determine the hot topic corresponding to the first hot topic content; based on the multiple keywords, the entity recognition result, the hot topic, the sentiment tendency and the confidence level of the corresponding sentiment tendency, and the hot topic summary, generate the second hot topic content.

[0024] In an optional implementation, the analysis of the second hotspot content to obtain hotspot analysis results further includes:

[0025] The multiple keywords in the second hot topic content and the multiple entities in the entity recognition results are respectively used as target keywords and target entities;

[0026] Obtain related trending content associated with the trending topic within a first preset time period on multiple platforms;

[0027] Statistically analyze the frequency and timing of occurrence of the target keywords and target entities within the related trending content;

[0028] Calculate the hotspot weight of the second hotspot content based on the frequency and time of occurrence;

[0029] Based on the aforementioned related trending content, the trending trend and first dissemination path of the second trending content are determined;

[0030] Based on the hot topic, the sentiment tendency, the hot topic weight, the hot topic trend, and the first dissemination path, determine the type of impact of the second hot topic content on the target enterprise.

[0031] In one optional implementation, the impact type of the second hot topic content on the target enterprise includes: strong marketing opportunity, weak marketing opportunity, temporary marketing opportunity, strong public relations risk, weak public relations risk, and neutral information.

[0032] In an optional implementation, based on the second trending topic and the target public relations strategy, public relations recommendations for the target company are generated, including:

[0033] Based on the second hot topic content and the target public relations strategy, a hot topic feature vector of the target company is generated;

[0034] From the configured public relations case library, match the public relations cases corresponding to the hot topic feature vectors; the public relations case library contains multiple public relations cases, the feature vectors corresponding to each public relations case, public relations measures, and public relations results;

[0035] Based on the public relations measures and results of the aforementioned public relations case, public relations recommendations are generated for the target company.

[0036] Secondly, the present invention provides an analysis device for enterprise hot topic information, the device comprising:

[0037] The acquisition unit is used to acquire the hotspot batch processing rules configured by the target enterprise; wherein, the hotspot batch processing rules include: hotspot search rules and hotspot content processing rules;

[0038] The data collection unit is used to collect first hot topics related to the target enterprise from multiple platforms based on the hot topic search rules.

[0039] The processing unit is used to process the first hot content based on the hot content processing rules to obtain the second hot content;

[0040] An analysis unit is used to analyze the second hot topic content and obtain hot topic analysis results; wherein, the hot topic analysis results include: the type of impact of the second hot topic content on the target enterprise;

[0041] The matching unit is used to match the target public relations strategy corresponding to the influence type from a configured table of different influence types and different public relations strategies;

[0042] The generation unit is used to generate public relations recommendations for the target company based on the second hot topic content and the target public relations strategy.

[0043] Thirdly, the present invention provides an electronic device, the electronic device including a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other through the communication bus;

[0044] Memory, used to store computer programs;

[0045] When a processor executes a program stored in memory, it implements the method described in any of the foregoing embodiments.

[0046] Fourthly, the present invention provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the method described in any of the foregoing embodiments.

[0047] This application separates content processing from the content library, improving system maintainability and processing efficiency, with data processing efficiency increased by more than 30% compared to existing technologies. This application can accurately and comprehensively extract key content points; for example, keyword extraction accuracy is improved by 10% compared to traditional algorithms, sentiment analysis accuracy reaches over 90%, and the F1 score for entity recognition is improved by 20%, providing users with more valuable content information. This application enables companies to collect and analyze trending content in a targeted manner to determine the type of impact of trending topics on the respective companies, allowing them to take timely public relations measures. Attached Figure Description

[0048] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0049] Figure 1 A flowchart illustrating a method for analyzing enterprise hotspot information provided in this application embodiment;

[0050] Figure 2 A schematic diagram of the structure of an enterprise hotspot information analysis device provided in this application embodiment;

[0051] Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0052] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0053] The method for analyzing enterprise hotspot information provided in this application can be applied to an enterprise hotspot information analysis system architecture. This system can include: an enterprise backend server and enterprise employee terminals. The server can be a physical server, a server cluster composed of multiple physical servers, or a distributed system. It can also be a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms. Terminals can be user equipment (UE) such as mobile phones, smartphones, laptops, digital radio receivers, personal digital assistants (PDAs), and tablet computers (PADs), handheld devices, in-vehicle devices, wearable devices, computing devices, or other processing devices connected to a wireless modem, mobile stations (MS), and mobile terminals. Terminals and servers can be directly or indirectly connected via wired or wireless communication methods; this application does not limit the connection.

[0054] In this embodiment, the server is equipped with an algorithm processing engine, an algorithm management platform, an Apache Kafka message queue, a distributed storage system, a backend service framework, Spring Cloud Alibaba, and a distributed system environment. The distributed storage system includes HDFS, Spark, and HBase for storing and processing massive amounts of data. The Apache Kafka message queue is used to implement asynchronous communication between internal system components and manage task queues. The algorithm processing engine is used to employ deep learning frameworks such as TensorFlow and corresponding algorithm models. The algorithm management platform is used to manage the uploading, deployment, and version control of algorithm models. Simultaneously, a permission management system and a data lineage tracking system are also deployed on the server to ensure system security and data traceability.

[0055] The preferred embodiments of this application are described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit this application. Furthermore, the embodiments and features in the embodiments of this application can be combined with each other without conflict.

[0056] Figure 1 This is a flowchart illustrating a method for analyzing trending information about an enterprise, as provided in an embodiment of this application. Figure 1 As shown, the method may include:

[0057] Step S110: Obtain the hotspot batch processing rules configured by the target enterprise.

[0058] The hot topic batch processing rules include: hot topic search rules, hot topic content processing rules, processing priority, hot topic type, hot topic filtering rules, key point strategies, hot topic strategies, plan execution count, root node input content, nested loop rules, and conditional branch processing logic. Hot topic search rules include: hot topic keywords, filtering rules, at least one search platform, and the corresponding search strategy for that platform. Hot topic keywords can be related to the company brand, products, executives, key projects, etc.; they can also be industry core concepts, competitors, technology trends, etc.; or they can be empty. When hot topic keywords are empty, the system automatically searches the hot lists and trending hashtags on major social media platforms, news websites, and forums. Processing priority is used to characterize the processing order, relevance threshold, response time, and resource investment of the hot topic batch processing plan. Users can configure hot topic batch processing rules through a graphical interface or code input. Hot topic content processing rules include: text cleaning standards, core domain constraints, word association priority, information extraction content, and keyword determination criteria.

[0059] Step S120: Based on hotspot search rules, collect the most popular content related to the target company from multiple platforms.

[0060] The most popular content can be long articles, text and images, videos, plain text, pure image galleries, pure videos, or pure audio, etc.

[0061] Specifically, for any search platform, based on the search platform's search strategy, initial search content related to hot keywords is retrieved from the search platform; the relevance between the initial search content and the hot keywords is calculated; if the relevance is greater than the configured relevance threshold, the initial search content is used as the first search content; the first search content of each search platform is sorted according to the relevance from largest to smallest; the sorted first search content is filtered using filtering rules to obtain the second search content; and the second search content is determined as the first hot content.

[0062] In some embodiments of this application, before calculating the correlation between the initial search content and hot keywords, the method further includes: performing text cleaning, multimodal recognition, and information extraction on the initial search content according to hot content processing rules; wherein, text cleaning includes: deduplication, noise reduction, and format unification; multimodal recognition is used to extract text from images, recognize speech-to-text in videos, and recognize objects or scenes in images / videos; information extraction is used to extract metadata such as publication time, source, author, and interaction data (likes, comments, and reposts).

[0063] Step S130: Based on the hot topic content processing rules, process the first hot topic content to obtain the second hot topic content.

[0064] Specifically, based on the rules for handling trending content, the top trending content is processed, including:

[0065] (1) Based on the hot topic content processing rules, the first hot topic content is segmented into multiple keywords using a pre-trained word graph model;

[0066] Specifically, the first hot topic content is segmented using a word graph model to obtain a first word sequence composed of multiple words. Based on the configured hot topic content processing rules, the initial word sequence is filtered to obtain a second word sequence. Using a pre-trained word graph model's word frequency dictionary and hot topic content processing rules, the words in the second word sequence are prioritized to obtain multiple keywords. The word graph model is constructed using the TextRank algorithm and contains multiple nodes and edges. Each node represents a word, and each edge represents the co-occurrence relationship between words. The word graph model continuously captures new industry terms and recent new product / strategic terms related to the target company's domain. Through the semantic correlation calculation between new and old words, it automatically iterates and updates the weights of the word graph nodes and edges until the node weights converge, selecting words with higher weights as keywords.

[0067] (2) Input the first hot topic content into a pre-trained BERT-based fine-tuned sentiment analysis model to obtain the sentiment tendency of the first hot topic content and the confidence level of the corresponding sentiment tendency; wherein, the BERT-based fine-tuned sentiment analysis model includes: an input layer, used to segment and encode the first hot topic content to obtain a tensor composed of word vectors, segment vectors and position vectors; a Transformer encoder stack: BERT-base contains 12 Transformer layers, BERT-large contains 24 layers, each layer is composed of a multi-head self-attention mechanism and a feedforward neural network; multi-head self-attention is used to capture the dependency relationship between different word vectors in the tensor; the feedforward neural network is used to perform non-linear transformation on the attention output to obtain the feature vector corresponding to each vector in the corresponding tensor; a classification head layer, used to classify based on BERT The output feature vectors are used for sentiment classification, outputting sentiment tendency and confidence score. A sentiment entity association layer is used to trace back through attention weights to locate the word vectors in the most popular content that trigger that sentiment tendency. The association strength between the triggering word vector and the corresponding sentiment tendency is calculated, generating a sentiment-word and association strength triplet, and the confidence score of the sentiment tendency is adjusted based on the association strength. A fully connected layer maps the feature vectors to sentiment tendencies. An activation function converts the output of the fully connected layer into a probability distribution, obtaining the confidence score corresponding to each sentiment tendency. The sentiment tendency is used to characterize whether the sentiment of popular content towards the company is positive, negative, or neutral.

[0068] (3) Input the first hotspot content into the pre-trained BiLSTM-CRF model to obtain the entity recognition result; wherein, the BiLSTM-CRF model includes: BiLSTM module and CRF module; the BiLSTM module is used to calculate the first hidden state and the second hidden state of the first hotspot content using forward LSTM and backward LSTM respectively; based on the first hidden state and the second hidden state, the fused hidden state is obtained; the fused hidden state integrates the context information; the CRF module is used to convert the fused hidden state into an entity label sequence, i.e., the entity recognition result; the entity recognition result includes specific products, people, events and places mentioned in the hotspot content;

[0069] (4) Input the obtained keywords, sentiment tendencies and corresponding sentiment tendency confidence scores, entity recognition results and first hot content into a pre-trained attention-enhanced sequence-based summary generation model to obtain a hot summary of the first hot content; the attention-enhanced sequence-based summary generation model includes a decoder and an encoder; the encoder is used to encode multiple keywords, sentiment tendencies and corresponding sentiment tendency confidence scores, entity recognition results and first hot content into a fixed-length vector, and the decoder is used to generate a hot summary based on the vector; during the generation process, the encoder calculates the attention weight of each position in the input multiple keywords, sentiment tendencies and corresponding sentiment tendency confidence scores, entity recognition results and first hot content with the current time; the decoder needs to refer to the attention weight when calculating the hot summary; the hot summary is used to represent the main evaluation or discussion focus of users on enterprises or enterprise products;

[0070] (5) Based on the hot topic summary, determine the hot topic corresponding to the first hot topic content; based on multiple keywords, entity recognition results, hot topic, sentiment-word and association strength triplet and corresponding sentiment confidence and hot topic summary, generate the second hot topic content.

[0071] Step S140: Analyze the second hot topic content to obtain the hot topic analysis results.

[0072] The hotspot analysis results include: the types of impact of the second hotspot content on the target company; the types of impact of the second hotspot content on the target company include: strong marketing opportunities, weak marketing opportunities, temporary marketing opportunities, strong public relations risks, weak public relations risks, and neutral information.

[0073] Specifically, multiple keywords from the second hot topic content and multiple entities from the entity recognition results are used as target keywords and target entities, respectively. Related hot topic content associated with the hot topic within a first preset time period on multiple platforms is obtained. The frequency and time of occurrence of target keywords and target entities in the related hot topic content are statistically analyzed. Based on the frequency and time of occurrence, the hot topic weight of the second hot topic content is calculated. Based on the related hot topic content, the hot topic trend and first propagation path of the second hot topic content are determined. Specifically, a breadth-first search algorithm based on graph theory is used to analyze the related hot topic content to obtain the first propagation path of the second hot topic content. Specifically, users and events on the platform are constructed into a directed graph, with user actions such as forwarding and commenting forming the edges of the graph. Starting from the starting node of the hot topic event, the BFS algorithm is used to traverse the graph, recording the path and scope of event propagation to obtain the first propagation path. Based on the hot topic, sentiment tendency, hot topic weight, hot topic trend, and first propagation path, the type of impact of the second hot topic content on the target enterprise is determined.

[0074] Step S150: Match the target public relations strategy corresponding to the different impact types from the configured table of different impact types and public relations strategies; generate public relations suggestions for the target company based on the second hot topic content and the target public relations strategy.

[0075] Among them, the public relations strategies corresponding to strong marketing opportunities, weak marketing opportunities, and temporary marketing opportunities are to amplify, spread, or accumulate value, specifically including increasing brand exposure, strengthening user goodwill, and converting into business growth; the public relations strategies corresponding to public relations risks and weak public relations risks are to cool down, stop losses, or repair trust, specifically including reducing the scope of dissemination, reducing user aversion, and salvaging the brand image; the public relations strategies corresponding to neutral information are to guide or convert, specifically including guiding neutral discussions to a positive direction or preventing them from evolving into a negative one.

[0076] Specifically, based on the second hot topic content and the target public relations strategy, a hot topic feature vector is generated for the target company; from the configured public relations case library, public relations cases corresponding to the hot topic feature vector are matched; the public relations case library contains multiple public relations cases, the feature vector corresponding to each public relations case, effective public relations measures, ineffective public relations measures, suitable channels, public relations results, and effect weights; based on the public relations measures and public relations results of the public relations cases, public relations suggestions for the target company are generated.

[0077] In some embodiments of this application, before generating public relations recommendations for the target company based on the second hot topic content and the target public relations strategy, the method further includes:

[0078] The process involves: acquiring trending topic data from different platforms within a second preset time period; preprocessing the acquired trending topic data to obtain first trending topic data; the second preset time period being the current time period, i.e., the latest trending topic data; calculating the first correlation between different first trending topic data; clustering the different first trending topic data based on the first correlation to obtain second trending topic data and clustering themes corresponding to multiple trending events; using the DBSCAN algorithm for clustering different first trending topic data; calculating the second correlation between the second trending topic data corresponding to any trending event and the target enterprise; inputting the second trending topic data corresponding to the trending event into a pre-trained trending topic popularity prediction model to obtain the popularity prediction result of the trending event; the trending topic popularity prediction model using the ARIMA time series prediction model; analyzing the second trending topic data corresponding to the trending event to obtain the second propagation path of the trending event; and generating trending features of the trending event based on the popularity prediction result, the second correlation, the second propagation path, and the clustering themes.

[0079] In the above embodiments of this application, the method for generating public relations recommendations for the target company may further include:

[0080] Calculate the semantic similarity of clustered themes in the hot topic features of hot events and the words in the sentiment-word and association strength triplet in the second hot content. If the similarity of semantic similarity is greater than the configured first similarity threshold, then based on the hot topic prediction result, the second propagation path, and the second relevance of the hot topic, and the corresponding sentiment-word and association strength triplet, generate hot topic sentiment features. Convert the hot topic sentiment features into standardized vectors, and use the cosine similarity algorithm to match historical public relations cases in the public relations case library with feature vector similarity greater than the second similarity threshold. Extract the effective public relations measures, appropriate channels, and effect weights of the historical public relations cases to generate case reference features. Construct a two-dimensional feature matrix with hot topic sentiment features as rows and case reference features as columns. The matrix elements are the adaptation scores of each hot topic feature and different public relations case measures (adaptation score = hot topic enterprise relevance × case effect weight × sentiment matching degree, sentiment matching degree: 1 if the hot topic and case sentiment are consistent, 0.3 if they are opposite, and 0.6 if neutral).

[0081] Sort the fit scores in the two-dimensional feature matrix in descending order, and take the top N historical case public relations measures as candidate reference measures; where N is a positive integer;

[0082] Based on the characteristics of the hot topics and the content of the second hot topic, the proposed reference measures are optimized to obtain the optimized proposed reference measures: Regarding the heat prediction results: if the heat prediction is rising, add a pre-implementation clause to the proposed reference measures (e.g., initiate measures 24 hours before the heat peak); if the heat prediction is falling, add a long-tail coverage clause (e.g., supplement content on platforms at the end of the dissemination path); regarding the second dissemination path: replace the implementation channel of the proposed reference measures with the platforms in the second dissemination path of the hot topic; regarding the sentiment-word and association strength triplet: if the sentiment is negative, add a word response clause to the proposed reference measures; if the sentiment is positive, add a word amplification clause, with higher association strength resulting in higher clause priority.

[0083] Based on the five elements of measures objectives, implementation actions, channels, time, and effect monitoring indicators, the optimized candidate reference measures are integrated into an initial public relations strategy;

[0084] Based on the second degree of relevance between trending events and the target company, the initial public relations strategy is optimized to obtain an optimized public relations strategy. If the relevance is greater than the first company relevance threshold, high-frequency monitoring clauses are added to the initial public relations strategy. If the relevance is less than the second company relevance threshold, cost control clauses are added to the initial public relations strategy. If the relevance is greater than the second company relevance threshold but less than the first company relevance threshold, the initial public relations strategy is not optimized. At the same time, the optimized public relations strategy is verified to be consistent with sentiment-words and relevance strength. If they are consistent, the optimized public relations strategy is converted into public relations suggestions in natural language.

[0085] Public relations recommendations include: Content creation suggestions: automatically generating content titles, summaries, and keyword suggestions based on trending topics and company business; Marketing campaign integration: suggesting the integration of trending elements into recent marketing campaigns or social media interactions; Partner collaboration: suggesting joint statements if the trending topic is related to partners; Crisis escalation warning: automatically notifying the public relations and legal teams; Response strategy suggestions: providing preliminary response points or statement draft frameworks based on historical cases and the current situation; Negative information dissemination path tracking: focusing on analyzing the dissemination nodes and key figures of negative information.

[0086] In some other embodiments of this application, the method further includes: establishing a crisis early warning model by combining negative emotions, specific crisis keywords (such as "failure", "complaint", "layoff"), and abnormal increase in interaction volume.

[0087] In other embodiments of this application, the method further includes: displaying the generated hotspot analysis results in a configured view format or a custom view template; the analysis view format includes: a timeline view, which allows users to view the timeline of content or hot events; a word cloud view, which visually displays the distribution and importance of keywords; and a regional heat map, which displays the popularity distribution of content or hot events in different regions; users can click on elements in the view to view more detailed data; when an element in a view is selected, the data in related views will also be updated accordingly.

[0088] In other embodiments of this application, the method further includes: obtaining hot data after implementing public relations recommendations, and comparing and analyzing the hot data after implementing public relations recommendations with the first hot content to obtain the public relations effect weight; generating new case data by combining hot feature vectors, public relations measures and public relations effect weights; if the effect indicators reach the preset standard, then classifying them into the case library according to the clustering theme; if the effect does not meet the standard, marking them as cases to be optimized, and prompting the user to supplement optimization measures.

[0089] Corresponding to the above method, embodiments of this application also provide an analysis device for enterprise hotspot information, such as... Figure 2 As shown, the device for analyzing the company's hot topics includes:

[0090] The acquisition unit 210 is used to acquire the hotspot batch processing rules configured by the target enterprise; wherein, the hotspot batch processing rules include: hotspot search rules and hotspot content processing rules;

[0091] The collection unit 220 is used to collect the most popular content related to the target company from multiple platforms based on hotspot search rules.

[0092] The processing unit 230 is used to process the first hot content based on the hot content processing rules to obtain the second hot content;

[0093] Analysis unit 240 is used to analyze the second hot topic content and obtain hot topic analysis results; wherein, the hot topic analysis results include: the type of impact of the second hot topic content on the target enterprise;

[0094] Matching unit 250 is used to match the target public relations strategy corresponding to the impact type from a configured table of different impact types and different public relations strategies;

[0095] Generation unit 260 is used to generate public relations suggestions for the target company based on the second hot topic content and the target public relations strategy.

[0096] The functions of each functional unit of the enterprise hotspot information analysis device provided in the above embodiments of this application can be implemented through the above methods and steps. Therefore, the specific working process and beneficial effects of each unit in the enterprise hotspot information analysis device provided in the embodiments of this application will not be repeated here.

[0097] This application also provides an electronic device, such as... Figure 3 As shown, it includes a processor 310, a communication interface 320, a memory 330, and a communication bus 340, wherein the processor 310, the communication interface 320, and the memory 330 communicate with each other through the communication bus 340.

[0098] Memory 330 is used to store computer programs;

[0099] When the processor 310 executes the program stored in the memory 330, it performs the following steps:

[0100] Retrieve the hot topic batch processing rules configured by the target enterprise; these rules include hot topic search rules and hot topic content processing rules.

[0101] Based on trending search rules, the most popular content related to the target company is collected from multiple platforms.

[0102] Based on the rules for processing trending content, the first trending content is processed to obtain the second trending content;

[0103] The second hot topic was analyzed to obtain the hot topic analysis results; among them, the hot topic analysis results include: the type of impact of the second hot topic on the target company;

[0104] Match the target public relations strategy corresponding to the different impact types from the configuration table of different impact types and different public relations strategies;

[0105] Based on the second hot topic and the target public relations strategy, generate public relations suggestions for the target company.

[0106] The communication bus mentioned above can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. This communication bus can be divided into address bus, data bus, control bus, etc. For ease of illustration, only one thick line is used to represent it in the diagram, but this does not mean that there is only one bus or one type of bus.

[0107] The communication interface is used for communication between the aforementioned electronic devices and other devices.

[0108] The memory may include random access memory (RAM) or non-volatile memory (NVM), such as at least one disk storage device. Optionally, the memory may also be at least one storage device located remotely from the aforementioned processor.

[0109] The processors mentioned above can be general-purpose processors, including central processing units (CPUs), network processors (NPs), etc.; they can also be digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0110] The implementation methods and beneficial effects of the various components of the electronic device in the above embodiments for solving the problem can be found in [reference needed]. Figure 1 The steps in the illustrated embodiments are used to implement the electronic device. Therefore, the specific working process and beneficial effects of the electronic device provided in this application will not be repeated here.

[0111] In another embodiment provided in this application, a computer-readable storage medium is also provided, which stores instructions that, when executed on a computer, cause the computer to perform the analysis method for enterprise hotspot information as described in any of the above embodiments.

[0112] In another embodiment provided in this application, a computer program product containing instructions is also provided, which, when run on a computer, causes the computer to execute the method for analyzing enterprise hotspot information as described in any of the above embodiments.

[0113] Those skilled in the art will understand that the embodiments in this application can be provided as methods, systems, or computer program products. Therefore, the embodiments in this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the embodiments in this application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0114] This application describes embodiments of methods, apparatus (systems), and computer program products according to embodiments of this application with reference to flowchart illustrations and / or block diagrams. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0115] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0116] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0117] Although preferred embodiments have been described in this application, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the embodiments of this application.

[0118] Obviously, those skilled in the art can make various modifications and variations to the embodiments of this application without departing from the spirit and scope of the embodiments of this application. Therefore, if these modifications and variations to the embodiments of this application fall within the scope of this application and its equivalents, then these modifications and variations are also intended to be included in the embodiments of this application.

Claims

1. A method for analyzing hot topics in enterprise information, characterized in that, The method includes: Obtain the hotspot batch processing rules configured by the target enterprise; wherein, the hotspot batch processing rules include: hotspot search rules and hotspot content processing rules; Based on the aforementioned hotspot search rules, first hotspot content related to the target company is collected from multiple platforms; Based on the aforementioned hot topic content processing rules, the first hot topic content is processed to obtain the second hot topic content; The second hot topic is analyzed to obtain hot topic analysis results; wherein, the hot topic analysis results include: the type of impact of the second hot topic on the target enterprise; Match the target public relations strategy corresponding to the different impact types from the configured table of different impact types and different public relations strategies; Based on the second hot topic and the target public relations strategy, public relations recommendations for the target company are generated. Generate public relations recommendations for the target company, including: The process involves: acquiring trending topic data from different platforms within a second preset time period; preprocessing the acquired trending topic data to obtain first trending topic data; clustering the different first trending topic data based on the first correlation between them to obtain second trending topic data and cluster themes corresponding to multiple trending events; calculating the second correlation between the second trending topic data corresponding to any trending event and the target enterprise; inputting the second trending topic data corresponding to the trending event into a pre-trained trending topic popularity prediction model to obtain the popularity prediction result of the trending event; and analyzing the second trending topic data corresponding to the trending event to obtain the second propagation path of the trending event and generate the trending features of the trending event. Calculate the semantic similarity of clustered topics in the hot topic features of the hot topic event and words in the sentiment-word and association strength triplet in the second hot topic content; if the similar semantic similarity is greater than the configured first similarity threshold, then based on the hot topic event's heat prediction result, the second propagation path, the second association degree, and the corresponding sentiment-word and association strength triplet, generate hot topic sentiment features, convert the hot topic sentiment features into standardized vectors, match historical public relations cases in the public relations case library whose feature vector similarity is greater than the second similarity threshold, extract the effective public relations measures, adaptation channels, and effect weights of the historical public relations cases, and generate case reference features; construct a two-dimensional feature matrix with hot topic sentiment features as rows and case reference features as columns, where the matrix elements are the adaptation scores of each hot topic feature and different public relations case measures; Sort the fit scores in the two-dimensional feature matrix in descending order, and take the top N historical case public relations measures as candidate reference measures; where N is a positive integer; Based on the characteristics of the hot topics and the content of the second hot topics, the selected reference measures are optimized to obtain the optimized public relations strategy; the optimized public relations strategy is then converted into public relations suggestions in natural language.

2. The method as described in claim 1, characterized in that, The hot search rules include: hot keywords, filtering rules, at least one search platform, and the search strategies of the corresponding search platforms.

3. The method as described in claim 2, characterized in that, The process of collecting first-trendy content related to the target company from multiple platforms based on the aforementioned hotspot search rules includes: For any search platform, based on the search strategy of the search platform, retrieve initial search content associated with the hot keywords from the search platform; Calculate the correlation between the initial search content and the hot keywords; if the correlation is greater than the configured correlation threshold, then use the initial search content as the first search content. Sort the top search results on each search platform in descending order of relevance. The sorted first search results are filtered using the aforementioned filtering rules to obtain the second search results; The second search result is identified as the top trending topic.

4. The method as described in claim 3, characterized in that, The process of processing the first hot topic content based on the hot topic content processing rules to obtain the second hot topic content includes: Based on the aforementioned hot topic content processing rules, the first hot topic content is segmented using a pre-trained word graph model to obtain multiple keywords; The first hot topic content is input into a pre-trained sentiment analysis model based on BERT fine-tuning to obtain the sentiment tendency of the first hot topic content and the confidence level of the corresponding sentiment tendency. The first hotspot content is input into the pre-trained BiLSTM-CRF model to obtain entity recognition results; The obtained keywords, sentiment tendencies and corresponding sentiment tendency confidence scores, entity recognition results, and the first hot topic content are input into a pre-trained attention-enhanced sequence-based summary generation model to obtain a hot topic summary of the first hot topic content. Based on the hot topic summary, determine the hot topic corresponding to the first hot topic content; based on the multiple keywords, the entity recognition result, the hot topic, the sentiment tendency and the confidence level of the corresponding sentiment tendency, and the hot topic summary, generate the second hot topic content.

5. The method as described in claim 4, characterized in that, The analysis of the second hot topic content yields the following hot topic analysis results: The multiple keywords in the second hot topic content and the multiple entities in the entity recognition results are respectively used as target keywords and target entities; Obtain related trending content associated with the trending topic within a first preset time period on multiple platforms; Statistically analyze the frequency and timing of occurrence of the target keywords and target entities within the related trending content; Calculate the hotspot weight of the second hotspot content based on the frequency and time of occurrence; Based on the aforementioned related trending content, the trending trend and first dissemination path of the second trending content are determined; Based on the hot topic, the sentiment tendency, the hot topic weight, the hot topic trend, and the first dissemination path, determine the type of impact of the second hot topic content on the target enterprise.

6. The method as described in claim 1, characterized in that, The impact types of the second hot topic content on the target company include: strong marketing opportunities, weak marketing opportunities, temporary marketing opportunities, strong public relations risks, weak public relations risks, and neutral information.

7. The method as described in claim 1, characterized in that, Based on the second hot topic and the target public relations strategy, public relations recommendations for the target company are generated, including: Based on the second hot topic content and the target public relations strategy, a hot topic feature vector of the target company is generated; From the configured public relations case library, match the public relations cases corresponding to the hot topic feature vectors; the public relations case library contains multiple public relations cases, the feature vectors corresponding to each public relations case, public relations measures, and public relations results; Based on the public relations measures and results of the aforementioned public relations case, public relations recommendations are generated for the target company.

8. A device for analyzing hot topic information of an enterprise, characterized in that, The device includes: The acquisition unit is used to acquire the hotspot batch processing rules configured by the target enterprise; wherein, the hotspot batch processing rules include: hotspot search rules and hotspot content processing rules; The data collection unit is used to collect first hot topics related to the target enterprise from multiple platforms based on the hot topic search rules. The processing unit is used to process the first hot content based on the hot content processing rules to obtain the second hot content; An analysis unit is used to analyze the second hot topic content and obtain hot topic analysis results; wherein, the hot topic analysis results include: the type of impact of the second hot topic content on the target enterprise; The matching unit is used to match the target public relations strategy corresponding to the influence type from a configured table of different influence types and different public relations strategies; The generation unit is used to generate public relations suggestions for the target company based on the second hot topic content and the target public relations strategy; Generate public relations recommendations for the target company, including: The process involves: acquiring trending topic data from different platforms within a second preset time period; preprocessing the acquired trending topic data to obtain first trending topic data; clustering the different first trending topic data based on the first correlation between them to obtain second trending topic data and cluster themes corresponding to multiple trending events; calculating the second correlation between the second trending topic data corresponding to any trending event and the target enterprise; inputting the second trending topic data corresponding to the trending event into a pre-trained trending topic popularity prediction model to obtain the popularity prediction result of the trending event; and analyzing the second trending topic data corresponding to the trending event to obtain the second propagation path of the trending event and generate the trending features of the trending event. Calculate the semantic similarity of clustered topics in the hot topic features of the hot topic event and words in the sentiment-word and association strength triplet in the second hot topic content; if the similar semantic similarity is greater than the configured first similarity threshold, then based on the hot topic event's heat prediction result, the second propagation path, the second association degree, and the corresponding sentiment-word and association strength triplet, generate hot topic sentiment features, convert the hot topic sentiment features into standardized vectors, match historical public relations cases in the public relations case library whose feature vector similarity is greater than the second similarity threshold, extract the effective public relations measures, adaptation channels, and effect weights of the historical public relations cases, and generate case reference features; construct a two-dimensional feature matrix with hot topic sentiment features as rows and case reference features as columns, where the matrix elements are the adaptation scores of each hot topic feature and different public relations case measures; Sort the fit scores in the two-dimensional feature matrix in descending order, and take the top N historical case public relations measures as candidate reference measures; where N is a positive integer; Based on the characteristics of the hot topics and the content of the second hot topics, the selected reference measures are optimized to obtain the optimized public relations strategy; the optimized public relations strategy is then converted into public relations suggestions in natural language.

9. An electronic device, characterized in that, The electronic device includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; Memory, used to store computer programs; A processor, when executing a program stored in memory, implements the method of any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the method described in any one of claims 1-7.

Images