77 results about "Topic type" patented technology

The invention provides a multi-role intelligent chatting method and system. The method comprises the following steps that information input by users and/or information subscribed by the users and/or information pushed to the users are/is acquired; semantics of the information are acquired through natural language processing, or the specific information is triggered according the preset condition; dialogue elements and at least two chatting robots participating in a dialogue are determined; a multi-role dialogue script is generated; dialogue lines are pushed to the chatting robots according to the dialogue script to achieve intelligent chatting. According to the method and system, through the mode that multiple chatting robots playing different roles have a group chat with the users, on one hand, the defect that in the current man-machine interaction process, the users need to continuously raise questions or a topic is overcome, and on the other hand, the topic is guided to be developed in a targeted mode through dialogue cooperation among the chatting robots according to the topic type and content, so that the interestingness, depth, breadth and knowledge transmission effect of information exchange are increased or enhanced. The invention further provides an information acquiring and information sharing method based on multi-role intelligent chatting. By means of role substation of the exclusive chatting robots of the users, the users who are not in the same time and space can communicate with one another just like they are face to face.

The invention discloses a question and answering (QA) system realization method based on deep learning and a topic model. The method comprises the steps of: S1, inputting a question sentence to the Twitter LDA topic model to obtain a topic type of the question sentence, extracting a corresponding topic word, and indicating the input question sentence and the topic word as word vectors; S2, inputting word vectors of the input question sentence to a recurrent neural network (RNN) for encoding to obtain an encoded hidden-layer state vector of the question sentence; S3, using a joint attention mechanism and combining local and global hybrid semantic vectors of the question sentence by a decoding recurrent neural network for decoding to generate words; S4, using a large-scale conversation corpus to train a deep-learning topic question and answering model based on an encoding-decoding framework; and S5, using the trained question and answering model to predict an answer to the input questionsentence, and generating answers related to a question sentence topic. The method makes up for the lack of exogenous knowledge of question and answering models, and increases richness and diversity of answers.

A computing system creates and stores semantic metadata on content, such as videos, that enables efficient searching of the content. The existing metadata of a video file, for example, is examined and a keyword list is created. The processes used to derive the keyword list may depend on the type and format of the existing metadata. The keywords from the list are compared against external structured knowledge data sources that are topic oriented. Based on these comparisons and the matches found, semantic data, including topic, topic type, and attribute data are inserted into a topic table. This uniform and structured table is stored on the computing system and is efficiently searchable for finding relevant videos and for finding relationships between videos.

The present invention relates to the technical field of text topic classification, and discloses a text topic classification method and system. According to the present invention, the text topic classification method comprises the following steps of: collecting corpuses, wherein the corpuses comprise texts of various types of topics; performing word segmentation on the corpuses, and performing text feature extraction on the corpuses after word segmentation to obtain feature vectors of the texts of the various types of topics; adjusting feature values in the feature vectors of the texts of the various types of topics according to a dynamic logarithmic excitation function, to obtain new feature vectors of the texts of the various types of topics; and according to similarity between a text to be classified and the new feature vectors of the texts of the various types of topics, classifying the text to be classified and determining a topic type of the text to be classified. In this way, text classification becomes more accurate.

The invention provides a method for automatically extracting key phrases of patent documents. The method includes steps of 1, preprocessing texts; 2, recognizing topic types of patented inventions; 3, extracting alternative key phrases and filtering the same; 4, subjecting the alternative key phrases to weight calculation and selecting the key phrases.

A computer-implemented method, computer-readable medium and system for scoring a text are disclosed. Themes within one or more texts may be determined and used to score each text, where an overall score for each text may indicate a respective importance and/or value of each text. The score for each text may be determined based upon a number of themes, type of themes, frequency of theme elements associated with the themes, distribution of theme elements associated with the themes, location of themes in the text, some combination thereof, etc. In this manner, the importance or value of one or more texts may be determined more accurately using information within each text with reduced reliance upon external information. Additionally, more relevant search results can be returned to a user by using internal information to perform ranking operations and/or filtering operations associated with a search.

The present invention helps generate connected communities of students and educators through access to an increased pool of granular learning resources that benefit from standardization of formats for creating and utilizing individual discussions on individual topics. The present invention also circumvents the complexity of existing repository systems by creating a standard platform format. Because this format is de facto implemented for any educator using the platform, standardized question types (e.g., discussion types or topic types) become functionally interchangeable notwithstanding any particular individuals utilizing the invention. The interchangeability of individual topics allows for creation of a global “topic library,” which may be made available to any educator using the system. This global topic library contains the standardized questions or discussion topics thereby allowing for quick review, selection, and incorporation into the individual teaching environment of any teacher utilizing the invention.

The present invention discloses a group message management method, comprising: obtaining content of group chat information and a user ID that sends the information; determining a topic type corresponding to the information according to a preset keyword; and adding information of a same topic type and a user ID that sends the corresponding information to a topic group corresponding to the topic type. The present invention further discloses a group message management apparatus. According to the group message management method and apparatus, group chat information may be filtered, and network resources and storage resources of a user terminal may be saved, which helps group chat communication become more efficient.

The invention provides an artificial intelligence-based article classification method and apparatus, a device and a readable medium. The method comprises the steps of obtaining texts of target articles; performing word segmentation processing of at least two different word segmentation granularities on the texts of the target articles to obtain segmented words corresponding to the word segmentation granularities; according to the segmented words, corresponding to the word segmentation granularities, of the target articles and pre-trained scoring tag models of all levels in a target classification system, predicting the similarities between the target articles and topic types on all the levels; and according to the similarities between the target articles and the topic types on all the levels, and a preset similarity threshold, classifying the target articles on all the levels. According to the technical scheme, the accuracy of classifying the articles is relatively high and the articles can be automatically classified, so that the time and labor are saved and the article classification efficiency is very high.

The invention relates to an Hbase database entering method for kafka. The method includes the steps that S1, all topic data in a kafka cluster is collected and stored in a queue; S2, the corresponding relation between topic and separators in the data, a filtering rule and a database entering rule are configured in a configuration file; S3, according to the corresponding relation, the data is serialized and filtered; S4, configuration information in the configuration file is read, connection with a Hbase database is set up, and the data is constructed into a put object; S5, the put object enters the Hbase. The problem that different topic types in kafka need to be processed individually is solved, the method general and adaptive to all topic methods is provided, the high-performance Hbase database entering method is constructed, the data writing efficiency is greatly improved, output bandwidth is increased, robot network and disk performance are used to the maximum, data safety is ensured to the maximum by means of double-queue design, and data loss is avoided.

The invention discloses a multi-media file recommending method and electronic device and relates to the technical field of multi-media. The multi-media file recommending method and electronic device can be used for recommending multi-media files preferred by users for different users, improving multi-media file recommending efficiency and enhancing user experience. According to the technical scheme, the method comprises the steps of acquiring preferred topic types of a present logging-in user from a multi-media client-side; confirming the multi-media file topic type of at least one multi-media file to be recommended and the affiliation granularity of different multi-media file topic types of the multi-media files to be recommended; calculating the recommending index of the multi-media file to be recommended according to the multi-media file topic type of the multi-media file to be recommended, the affiliation granularity of different multi-media file topic types of the multi-media file to be recommended and the preferred topic type; generating a multi-media file recommending list according to the recommending index of the at least one multi-media file to be recommended, and displaying the at least one multi-media file in the multi-media file recommending list.

Systems and methods provide a website builder tool capable of searching an image database to provide one or more proposed items of media for inclusion in a web page. The proposed media may correspond to one or more of the following: A keyword within the text input by the user for the web page; A language of the text; A theme genre assigned to the web page; or A geographical region associated with the web page. The proposed media can be presented to a user via an media selection user interface for selection as web page media for inclusion in the web page.

The invention discloses a text analysis method and a text analysis device and belongs to the field of information retrieval. The method comprises the steps of obtaining one or a plurality of kinds of characteristic information of a target text; respectively calculating a quantitative score of each kind of characteristic information of the target text; multiplying the quantitative score of each kind of characteristic information of the target text by respective corresponding weight and then accumulating each quantitative score to obtain the total scores of the target text. Since the quantitative scores of all kinds of characteristic information of the target text are respectively calculated, each quantitative score is accumulated according to the respective corresponding weight to obtain final scores and the characteristic weight and the effect of adaptive extension of characteristic items can be adaptively corrected in application scenarios, the problem that there is no proper method in the prior art for performing quality analysis on topics posted in forums is solved, and the effect of accurately analyzing the text quality of topic-type texts in the forums can be achieved.

The invention relates to a microblog hot topic detection method and device aiming at a specific group. The detection method comprises the following steps of performing community classifying on all users in the microblog according to the attention relation between the certificated users and the users concerned by the certificated users; extracting a topic according to the microblog content issued by each obtained certificated user and the users concerned by the certificated users so as to obtain multiple summary topic types; establishing a topic matrix according to the user communities and the summary topic types; placing a microblog text corresponding to the micorblog text vector in each summary topic type with the maximum cosine similarity with the microblog text vector; selecting a plurality of summary topic types with the maximum matrix element numbers in the community topic matrix for clustering and classifying respectively; respectively extracting central hot event of each event clustering type and showing the central hot spot event for the corresponding community users. A multilayered result of the hot topic event is obtained, and the hot event interesting to a certain community is represented by the microblog.

The invention discloses a human-computer interaction system for a service robot. The human-computer interaction system can be mounted on a universal movement platform including, but not limited to, intelligent terminals such as a tablet personal computer. The human-computer interaction system is characterized by dividing the domestic service robot system into a three-layer cloud infrastructure including an executing and sensing layer, an information center layer and an application source layer; packaging equipment as service provides by defining message files and topic types of the equipment; maintaining a resource list including an equipment list, a topic queue and a message queue in the information center layer, thus the equipment, such as the robot, serving as cloud service is provided for the user. In this way, multiple high-friendliness human-computer interaction manners, such as voice recognition, are facilitated through the cloud resource including voice cloud and image cloud. By the use of the human-computer interaction system, a three-dimensional visual simulation environment driven by real data is provided for the user, thus simulation demo is allowed, and the real system can be connected to control the robot and other equipment.

The invention discloses a recommendation method and network equipment. Through a topic model algorithm, a topic model of an object to be recommended is established, the topic model includes a topic type of the object to be recommended and the weight of the object to be recommended under the topic type, and the topic type includes at least two elements with semantic association relations; the similarity between the topic model of the object to be recommended and a preference feature model of a user is calculated, wherein the preference feature model of the user is obtained by calculation on the basis of historical operation records of the user under a platform to which the object to be recommended belongs; when the similarity between the topic model of the object to be recommended and the preference feature model of the user reaches a preset threshold value, the object to be recommended is recommended to the user. The problem is solved that currently, when recommendation is conducted on the basis of keywords, it is difficult for extracted keywords to accurately and comprehensively reflect behavioral preferences of the user, the requirements of the user can be more comprehensively met, the preferences of the user can be more comprehensively satisfied, and the user experience is improved.

The invention provides a network information processing method and device. The network information processing method mainly comprises the steps that a server obtains an assigned time period and a topic type selected by a user; the server obtains at least one topic corresponding to the assigned time period and the topic type, and determines at least one hot topic according to the number of messages left by the user in the topic or the number of pictures or videos contained in the messages lefty by the user. By means of the network information processing method and device, the user can conveniently obtain hot topic information corresponding to the assigned time period and the topic type selected by the user, and hot topics concerned by the user, particularly in a vertical field, can be presented immediately.

The invention discloses an electric system real-time data processing method. The method includes the steps of conducting classification on real-time data in an electric system according to a plurality of preset topic types, generating data files corresponding to the plurality of topic types according to the classified real-time data, storing the data files, receiving a data access request sent by a data access request end, and returning the data files corresponding to the topic types to the data access request end. According to the method, when a real-time data service access demand is generated inside the electric system, no data access interface needs to be specially opened, the access burden of an electric system database is greatly lightened, and the stability of electric system operation is enhanced. According to the electric system real-time data processing method, the supporting capacity of the electric system for real-time data access is greatly improved, meanwhile, the electric system real-time data processing method is easy to achieve, and the cost of system development is reduced.

A "Facet Recommender" creates conversational recommendations for facets of particular conversational topics, and optionally for things associated with those facets, from consumer reviews or other social media content. The Facet Recommender applies a machine-learned facet model and optional sentiment-model, to identify facets associated with spans or segments of the content and to determine neutral, positive, or negative consumer sentiment associated with those facets and, optionally, things associated with those facets. These facets are selected by the facet model from a list or set of manually defined or machine-learned facets for particular conversational topic types. The Facet Recommender then generates new conversational utterances (i.e., short neutral, positive or negative suggestions) about particular facets based on the sentiments associated with those facets. In various implementations, utterances are fit to one or more predefined conversational frameworks. Further, responses or suggestions provided as utterances may be personalized to individual users.

The invention relates to the technical field of information retrieval, and provides a semantic document retrieval method for searching topic type query in TED speech. The method comprises: training aneural network model by utilizing the existing query and document, and learning neural network model parameters; when a user inputs a query, a query likelihood retrieval model is used to obtain a preliminary retrieval result; and inputting the preliminary retrieval result into the neural network model with the fixed parameters for reordering, and determining a final retrieval result. According tothe method, the problem that a good effect cannot be achieved due to the fact that a traditional retrieval method in topic type query retrieval lacks semantic contact between queries and documents issolved; respectively modeling the topic type query and the speech document by introducing a neural network, and obtaining the semantic level correlation between the query and the document. In the neural network part, a recurrent neural network and a convolutional neural network are connected in series, and in addition, in order to solve the problem of gradient disappearance, a currently popular LSTM module is adopted.

A method, system and computer program for annotating a text for a user comprising a step executed by a developer to create a Topic Map summarizingthe topics of interest of the user; the developer further creating a data structure, a Topic class for each topic; an Annotator program executing on the computer automatically using the Topic Map to create two linked user dictionaries, a Topic Dictionaryreflecting Topic names and a Traversal Dictionary, preferably FSA-based dictionaries, reflecting the knowledge structure of a topic; the developer optional linking actions with topic types;the Annotator automatically instanciatingthe topic data structure of the topics found in the text using the dictionariesand actions when they have been previously linked to topic types;; the Annotator automatically attaching the instanciatedTopic data structure to the text; the developer creating a GUI allowing the user to access and interact with the text annotations.

The invention relates to a dynamic image output system for network electronic map and method for making the same. The invention merges and converts maps of different subject types into image files, and stores the image files in database for user accessing needed electronic map file directly. The dynamic image output system comprises a client terminal unit, a database unit, a website service unit and a network transmission unit. The invention has the advantages of being stable, of being able to lower server load and make plural users be served at the same time.

The invention discloses a data configuration based oil and gas resource data integration method and integration platform. Multi-source isomerous database and related space database are effectively organized through configuration files to provide various inquiring, positioning and editing functions for users and then data product are produced and displayed, so that dynamic, flexible and complete management of oil and gas resource data is achieved and various user requirements are satisfied, disadvantages caused by an information isolated island are overcame, a special topic type database which can accommodate various file formats and different data types is formed, supports are provided for integration processing and data mining of the oil and gas resource information, required data, information, background materials and policy materials are rapidly and accurately provided, an oil and gas resource strategic survey result is timely cleared up, and oil and gas resource result application services are developed.

The invention provides a method and a device for correcting oral arithmetic questions in an examination paper. A paper to be search is processed by a dictation engine. At the same time, according to the text content of the stem of each subject to be searched, the eigenvector of the subject to be searched is obtained, and then searching a target test paper matched with the test paper to be searchedfrom the test bank by utilizing the eigenvectors of each item to be searched, while for the searched questions whose topic type is oral arithmetic, based on the eigenvectors of the items in the target paper. The criterion for finding is the shortest editing distance the smallest, If the title type of the matched target title is also an oral arithmetic title, it is confirmed that the title to be searched is an open-ended arithmetic title, and the calculation result corresponding to the open-ended arithmetic title area of the oral arithmetic engine is used as the answer to the open-ended arithmetic title area, and the calculation result of the oral arithmetic engine outside the open-ended arithmetic title area of the open-ended arithmetic title area is discarded. The scheme provided by theinvention can improve the searching accuracy of the oral arithmetic problem.