118 results about "Topic analysis" patented technology

A system for analyzing vertical public opinions based on an industry comprises an acquisition and pre-treatment module for acquiring and pre-treating Internet information relevant to the consumer electronics industry and obtaining the formative information of the consumer electronics industry based on documents; a word segmentation module for matching words by means of a character string matching algorithm, and obtaining work segmentation results by amending the matching results in a word segmentation method based on understanding and statistics; an analysis module for performing document clustering and classification on the word segmentation results according to the frequency and similarity of keywords in the word segmentation results of the documents, and for obtaining analyzed and processed information after hotspot / sensitive topic analysis, orientation analysis and trend analysis to the clustered and classified results; and a display module for pushing the analyzed and processed information to users. The invention further provides a method for analyzing vertical public opinions based on an industry.

The invention discloses a microblog hot topic analyzing method. The method comprises the following steps that a microblog collection module obtains microblog data in the mode of combination of a web spider and a microblog third-party api technology according to a collection strategy; key words and sensitive words are called from a word bank through a word segmentation technology, and key words and sensitive words are analyzed out from microblog text data; the microblog webpage text data are filtered according to the analyzed key words, the analyzed sensitive words and emotional tendency words; a hot topic module marks the content involved between the symbols of # and # and between the symbols of [] as a topic through a clustering analysis technology, so that the number of microblog comments is counted; a hot people module analyzes the number of microblog fans and the number of the comments through the clustering analysis technology; a microblog early warning module analyzes out microblog information related to the key words and the sensitive words from the network microblog; an analyzing and counting module automatically generates a brief report through relevant data analyzed out from the system. The accuracy of topic analysis is improved, and detection efficiency is improved.

A topic analyzing method is provided in which the number of main topics in text data which is added in time series and generation and disappearance of topics are identified in real time as needed, and features of main topics are extracted and thereby one can know a change in the content of a topic with a minimum amount of memory and processing time. There is provided a system that detects topics while sequentially reading text data in a situation where the text data is added in time series, including learning means for representing a topic generation model by a mixture distribution model and learning the topic generation model online while more-heavily discounting the older data on the basis of a timestamp of the data; and model selecting means for selecting an optimal topic generation model from among a plurality of candidate topic generation models on the basis of information criteria of the topic generation models, wherein the topics are detected as mixture components of the optimal generation model.

The invention relates to a method and device for identifying a commodity with labels and a method for commodity navigation. The method for identifying the commodity with the labels includes the following steps that description information of the commodity is extracted; the description information of the commodity is converged to generate a text; a text analysis method based on a topic model is used for performing topic analysis on the text, multiple topics are obtained, and topic names are defined; the topic names related to the description information of the commodity serve as the labels of the commodity to identify the commodity. According to the method and device for identifying the commodity with the labels and the method for commodity navigation, the commodity can be identified with the labels with user dimensionality attributes, and therefore a user can conveniently, visually and fast find needed commodities.

The embodiment of the invention relates to a processing system for teaching data. The processing system comprises an educational terminal, a learning terminal and a shared platform, wherein the sharedplatform comprises a teaching live broadcasting module, a teaching recording module, a teaching on-demand broadcasting module and an online evaluation module; the teaching live broadcasting module comprises a live broadcasting establishing unit, a network connection unit, a video and audio receiving unit, a live broadcasting interaction unit and a content distributing unit; the teaching recordingmodule comprises a recording unit, an analyzing unit, an annotating unit and a storage unit; the teaching on-demand broadcasting module comprises an on-demand broadcasting receiving unit, an on-demand broadcasting querying unit and an on-demand broadcasting processing unit; the online evaluation module comprises a test question entering unit, a proofreading unit, a wrong topic analysis unit and aknowledge point pushing unit. When students view analysis for wrong questions tested on the platform, relevant explaining videos can be obtained according to knowledge points of the wrong questions and are pushed to the students to help the students to break through weak knowledge points and improve the learning efficiency; in addition, in the live broadcasting process, a teacher can monitor thelearning states of the students so as to guarantee the learning quality of the students and the teaching quality of the teacher.

The invention provides a label automatic generating method and system, a computer readable storage medium and equipment. The label automatic generating method comprises the steps of establishing an initial label set aiming at a training text with a label and a text with a to-be-generated label; performing mining on the training text with the label and the text with the to-be-generated label; training a label judging model; and according to the label judging model, searching a text label corresponding to the text with the to-be-generated label. According to the invention, the text analysis technology, machine learning and the deep learning algorithm are adopted, and information mining is carried out on the text data to be labeled on the basis of the original label set constructed by the multiple methods; based on the text topic analysis method, the distribution situation of words in the text is combined, so that similarity calculation of the text label theme of the multi-model fusion isrealized, the problems that text data such as internet online content are not labeled, and the labels are few are solved, and the problems that manual labeling lacks a unified standard, and differentusers can mark similar texts as different labels can be solved. Finally, a user can obtain expected information more accurately and more efficiently.

The invention relates to a method and a system for automatically calculating the evolutional trend of a topic on the internet. The prior art only can simply analyze the topic (or event) from a document in a centralization way, and give out the document information contained by the topic. In fact, each topic changes with the variation of the time, and the topic evolutes constantly on the time dimension. On the basis of the prior topic detection system, the invention periodically calculates the relation between the topic in the current period and the topic in the last period and stores the relations. The system takes out the relations between the topic information which is corresponded to a plurality of periods and the topics, and can visually display the evolutional trend of the topic over time at a client in a graphic mode according to the time range input by the user. By adopting the method of the invention, a more three-dimensional topic analysis result can be provided for the user, and the understanding and the recognition of the user to the topic are deepened, thereby helping the user to make a decision. The method is widely applicable to the intelligent information processing.

The invention discloses a financial behavior analyzing system based on social media calculation. The financial behavior analyzing system is characterized by comprising three modules of reptiles, databases and indexes, and an analyzer. The reptiles are in charge of acquiring data. The databases are divided into two parts of structured data and unstructured data. A global ID is set from each user and each microblog according to acquired data information when the indexes are established so as to perform align and retrieval on information in different databases. The analyzer is the core of the system and comprises six sub-modules of topic analysis, entity recognition, gesture recognition, message tracking, sentiment analysis and community cluster analysis. By the aid of the economic financial behavior analyzing system based on the social media calculation, user information can be acquired effectively and accurately, so that user data can be archived and arranged completely, user databases can be established, and information push which users concerned about can be provided to the users according to the user databases.

The invention discloses a text topic analysis technology based short text recommendation method. Information forwarded or published by a user is subjected to topic analysis by utilizing a text topic model to obtain topic preferences of the user, and information meeting the user preferences is recommended from large amounts of unread information, so that the information overloading problem of a system is better solved. Based on a biterm topic model (BTM) and a short text based aggregation method, a new short text topic analysis-oriented topic model, namely, a user-based biterm topic model (UBTM), is proposed; and an experiment in a real data set from microblog shows that the UBTM can obtain a topic with higher quality in comparison with a conventional short text topic analysis method. A UBTM based short text recommendation experiment also shows that the short text recommendation method proposed by the invention has a better recommendation effect.

The invention discloses an evaluating method for the product comment quality in electronic commerce. The evaluating method includes the steps that comment data is grabbed to construct a product comment document; the incidence relationship among product categories, themes and characteristic words contained by the themes is built with a theme analysis model; virtual concept lattices with the product categories as objects and the themes as properties are constructed with a formal concept analysis model; a comment-quality evaluating model is constructed; the comment data is obtained and subjected to word dividing operation; the divided words are input into the comment-quality evaluating model to conduct quality evaluation of the comment data; the quality evaluation result is output. By means of the evaluating method, the evaluation result of the product comment quality is recommended to a user in an ascending-order mode, and shopping decision of the user can be more objectively assisted. The relativity, the comprehensiveness, the detail performance and the professional performance of a product are evaluated and commented through four quantitative indexes, and the commented quality evaluation result can be obtained and provided for the user to refer.

The invention discloses a theme community discovery method based on a social network. The method includes the steps that 1, theme analysis is performed on document sets of the social network so as to obtain theme vector sets; 2, clustering is conducted on the theme vector sets through a k-means algorithm so as to obtain theme clusters; 3, link partition is conducted on each theme cluster to obtain a theme community set of each theme cluster. The theme community discovery method which can effectively and efficiently perform theme and link partition on communities is provided through combination of a community discovery algorithm based on link and a theme model algorithm.

The invention relates to a personalized travel package recommendation method based on demand classification and subject analysis. The method includes: analyzing natural language form demands input by a user, utilizing word segmentation, demand classification and other natural language processing techniques to process and classify the user demands so as to obtain rigid demand, flexible demand and negative demand of the user; then utilizing an LDA (latent dirichlet allocation) document theme generation model, making travel service individuals effectively cluster into different service fields by theme similarity, and then conducting similarity matching with the user demands so as to obtain a service list best matching user expectation; finally carrying out travel package design recommendation by means of travel package optimized recommendation algorithm: firstly acquiring a travel package scenic spot set according to user time demand and service priority information; then combining location information, preference information and the like to determine travel package hotel service; and then calculating the optimal journey of every day according to the distance function L, and ranking the travel package according to travel package recommendation index; and finally, selecting travel package catering service according to scenic spot location and user preference. By integrating the processing, the purpose of designing and recommending personalized travel package best meeting the user demand can be realized.

The invention discloses a Chinese commentary analysis method applicable to collecting a Chinese 'pseudo comment' corpus by analyzing Chinese comments of users to determine whether the comments can be used as corpus or not. The method includes the steps that a user submits a comment to a website, the front-end of the website sends an analysis request to a control center, the control center transfers the comment to an analysis unit, the analysis unit performs theme classification analysis on the comment, a word classification server performs word classification and part-of-speech tagging, the analysis unit sequentially performs syntactic analysis and sentiment analysis, and a data center saves analysis conclusions to a user comment table. According to the Chinese commentary analysis method, the control center can directly exclude unqualified corpus through thematic analysis and the analysis unit sequentially performs syntactic analysis and sentiment analysis on the comment of the user so as to effectively draw conclusions of emotional tendencies of the Chinese comment and improve the accuracy of an analysis system, and then an administrator can only view the comment in positive tendencies to determine whether the comment meets the requirements or not.

The invention discloses an economic and financial behavior analysis system model based on social media. The economic and financial behavior analysis system model based on the social media is characterized in that the system comprises a crawler, a database / indexer and an analyzer, wherein the crawler is mainly in charge of data collection; the database is divided into two parts which are structural data and unstructured data, when indexes are built, each user and each piece of microblog are respectively configured with a global ID according to collected data information, and information in different databases is aligned and searched; the analyzer is the core of the system and comprises a topic analysis submodule, an entity identification submodule, an action recognition submodule, a message tracking submodule, an emotion analysis submodule and a community cluster analysis submodule. The economic and financial behavior analysis system model based on the social media can effectively and accurately collect user information, conducts relatively complete archiving and arrangement on user data, builds a user information base, and provides the users with push of messages which are concerns of the users according to the user information base.

The invention discloses a topic analysis method, which comprises the steps of obtaining to-be-processed text corpora, and obtaining a word segmentation result and a corresponding part-of-speech corresponding to each to-be-processed text corpus; obtaining a filtered text corpus; analyzing the word segmentation result and the corresponding part-of-speech of each filtered text corpus through a dependency syntax to obtain a dependency relationship between syntax components of segmented words and the segmented words and a dependency pair corresponding to each text corpus; obtaining a topic corresponding to each text corpus according to the combined sentence pattern structure and the dependency pair; obtaining similar topics, and sorting the similar topics according to the number of the similartopics. The invention also discloses a topic analysis device and a storage medium, syntactic analysis is used on the basis of word segmentation to analyze a dependency relationship between a grammatical structure and a word segmentation result in a text statement, and a smooth and accurate topic is extracted according to a plurality of preset common Chinese combined sentence pattern structures, sothat topics can be analyzed from massive texts.

A social advertising facing Twitter feasibility analysis method includes the steps of building a multi-source Twitter corpus by innovatively combining corpus information of different sources of Twitter users and effectively expanding Twitter short text to infer the potential advertising value of the content published by the users to further achieve precise advertising audience targeting; proposing a multi-source Twitter corpus theme analysis model for latent semantic analysis of the content published by the users; based on semantic analysis results, designing feature selection, filtering and presentation algorithms, constructing a logistic regression classifier, and classifying advertising feasibility used as the basis for decision making of advertising recommendation. The social advertising facing Twitter feasibility analysis method takes full advantage of characteristics of information published by the users and can accurately infer the potential advertising value. By means of the social advertising facing Twitter feasibility analysis method, inferred results conforming to the intent of the users can be obtained. The social advertising facing Twitter feasibility analysis method is applicable to advertising recommendation of social networking services, such as Twitter.

The invention relates to a news data-based individual share emotion convergence method. The method comprises the following steps of: 1, crawling news information to form news documents and storing the news documents into a document storage database; 2, calculating the heat of each document and removing the repeated documents; 3, preprocessing content items in the news documents to form text sets; 4, carrying out comprehensive emotion analysis and theme analysis on each text set to form a two-tuple set, and carrying out text theme clustering and grouping; 5, integrating all the related financial reports to form an individual share-based triple set; 6, converging above results by taking individual shares as center; and 7, selecting a visible system to display the results to users. According to the method, correct and high-readability simplified theme emotion information can be provided for investors of financial markets, the investors can be helped to understand and better carry out investment judgement through shorter time, and important prediction model auxiliary information can be provided for quantifying the fund companies.

In a device, a generating unit generates the document-information sets corresponding to each joint-author from the relevant-document information set, an analysis-result storing unit stores the result of the user's analyzing the document-information sets corresponding to each joint-author in the analysis-result DB as a topic, an analysis-result integrating unit integrates the joint author's topics stored in the analysis-result DB and displays the topics integrated as the integrated analysis-result screen. Appropriate screens are displayed to support selection of the target for the analysis using the document-information sets that the generating unit generates corresponding to each joint-author.

In one embodiment of the present invention, a semantic analyzer translates a text segment into a structured representation that conveys the meaning of the text segment. Notably, the semantic analyzer leverages a semantic network to perform word sense disambiguation operations that map text words included in the text segment into concepts—word senses with a single, specific meaning—that are interconnected with relevance ratings. A topic generator then creates topics on-the-fly that includes one or more mapped concepts that are related within the context of the text segment. In this fashion, the topic generator tailors the semantic network to the text segment. A topic analyzer processes this tailored semantic network, generating a relevance-ranked list of topics as a meaningful proxy for the text segment. Advantageously, operating at the level of concepts and topics reduces the misinterpretations attributable to key word and statistical analysis methods.

The invention provides an enterprise classification method and system based on big data deep learning and electronic equipment, and the method comprises the steps: obtaining the comprehensive information of an enterprise, and forming a big data set; based on a CRF word segmentation model and a probability graph model, extracting an enterprise component keyword set, training a corresponding word vector model, and predicting and dividing a plurality of feature keyword sets by using a density clustering algorithm; carrying out TFI-DF screening on the word sets by utilizing a FastText text classification model, carrying out topic analysis on the big data set by utilizing an LDA model, extracting subject terms related to enterprises, and constructing a plurality of subject term sets by utilizing a density clustering algorithm; combining the feature keyword set and the subject term set to obtain a plurality of training samples, inputting the training samples into a bidirectional cycle neural network for training, and constructing a multi-category classification model; and carrying out classification prediction on enterprises by utilizing the multi-category classification model, matching a perfect threshold value, and automatically labeling industry labels of multiple hierarchies. The method has the characteristics of strong scene adaptability, high classification accuracy, high efficiency and reduced labor cost.

The invention relates to a news topic analysis method and a system for implementing the method. The method comprises the steps of (1) information collection and denoising step, (2) text information pre-processing step, (3) text information depth processing step, and (4) a step of constructing and displaying a relational graph of interested parties. Through the method and the system, the topic graph the relational network of the interested parties in the back of a news report can be deeply explored, and the reputation management can be performed better.

Described is a system for detecting attacks on networks. A hierarchical representation of activity of a communication network is used to detect and predict sources of misinformation in the communication network. The hierarchical representation includes temporal patterns of communication between at least one pair of nodes, each temporal pattern representing a motif, having a size, in the hierarchical representation. Changes in motifs provide a signal for a misinformation attack.

The invention relates to the technical field of government affair inquiry, in particular to a system for generating a six-dimensional knowledge graph, which comprises a knowledge base, a domain recognition module, an intention recognition module, a slot filling module, a similarity calculation module and an evaluation module, wherein the knowledge base is semi-structured and structured data from government sites and various vertical sites in the fields of livelihood, enterprises and affairs; the domain identification module is used for identifying questions consulted by users and dividing corresponding government affair domains; in the aspect of government affair information service, a government website intelligent search system and an intelligent question-answering system provide interactive services such as the folk affair handling field, policy consultation and complaint suggestion. The knowledge graph is the core basic capability of the AI, and provides basic data support for government affair knowledge base construction, such as government portal website knowledge base, AI artificial intelligence + government affair level, government department topic analysis and decision research, public opinion monitoring and the like.

The invention discloses a text label recommendation method based on a supervision topic model. According to the method, the new supervision text topic model Sim2Word is proposed by taking into account the characteristics that tags and related words frequently appear in corresponding texts, so that the problems that text keyword extraction methods have low prediction efficiency, and text topic analysis methods have low prediction accuracy are solved. The method includes the two main steps of firstly, acquiring the related word data of the existing tags based on a word vector technology, and then using the tags and the related words to train a tag prediction model; finally predicting the tags of new texts based on the model. Experiments on real datasets such as StackOverflow show that the method has higher recognition accuracy compared with traditional text tag recommendation techniques.

The invention provides a project domain theme analysis system based on a big data analysis technology. The system comprises a server and a user terminal. The server comprises a storage layer, an application layer and a communication layer. The storage layer comprises a project domain data warehouse module and a market module. The project domain data warehouse module is a data storage area for centralizing and integrating project domain historical data of an enterprise. The market module is used for acquiring different data sets for different secondary theme domains or different classifications. The application layer comprises an analysis module and an input module. The analysis module performs different types of topic analysis on the project domain historical data. The input module is usedfor inputting new data into the storage layer. The communication layer comprises a communication module and is used for establishing communication connection with a user terminal, and the user terminal is used for sending the input project domain related data to the server and obtaining a theme analysis result sent by the server. The system can meet the requirements of enterprises for comprehensive analysis and management of projects.

The invention discloses a dialog flow extraction method, a device and a storage medium based on intention analysis and dialog clustering. The method comprises the following steps of: acquiring original chat corpus, and carrying out subject analysis on sentences in the corpus through LDA algorithm, wherein the analyzed subject is called intention; screening confirms the topic as meaningful, and labels the sentences with valid intention in the corpus; the intention tags in the dialogue are extracted to form a sequence, which is called dialog flow. Further, the KNN clustering algorithm is adoptedfor all dialog flows to obtain k clusters and k dialog flows. The invention greatly reduces manual participation process and improves efficiency.

The invention provides a user topic participation prediction method, and belongs to the field of data mining and information retrieval. A data acquisition module acquires user information under a hot topic; a feature extraction module finds out an information triadic group formed by users participating in the topic of each time period by performing time slicing on the behavior of topic participation of the users, extracts feature properties for each user and extracts the properties of the information triadic group based on the properties of the users; a model training module performs modeling of the closing behavior of the information triadic group based on the properties of the information triadic group to construct a triadic information factor graph model and finds out the closed information triadic groups in the next stage of the hot topic; and a result prediction module predicts the users participating in the topic according to the predicted closing result of the information triadic groups. According to the method, the behavior of the users of participating in the topic is regarded as the closing behavior of the information triadic group so that a new idea is provided for topic participation prediction in the social network, and the method can be widely applied to the related fields of topic recommendation and topic analysis and the like.