33 results about "Concept network" patented technology

Search results are processed using search requests, including analyzing received queries in order to provide a more sophisticated understanding of the information being sought. A concept network is generated from a set of queries by parsing the queries into units and defining various relationships between the units. From these concept networks, queries can be automatically categorized into categories, or more generally, can be associated with one or more nodes of a taxonomy. The categorization can be used to alter the search results or the presentation of the results to the user. As an example of alterations of search results or presentation, the presentation might include a list of “suggestions” for related search query terms. As other examples, the corpus searched might vary depending on the category or the ordering or selection of the results to present to the user might vary depending on the category. Categorization might be done using a learned set of query-node pairs where a pair maps a particular query to a particular node in the taxonomy. The learned set might be initialized from a manual indication of which queries go with which nodes and enhanced has more searches are performed. One method of enhancement involves tracking post-query click activity to identify how a category estimate of a query might have varied from an actual category for the query as evidenced by the category of the post-query click activity, e.g., a particular hits of the search results that the user selected following the query. Another method involved determining relationships between units in the form of clusters and using clustering to modify the query-node pairs.

Multiple concept networks are generated from subsets of received queries. These concept networks can be used in various ways to enhance response to subsequent queries. In one embodiment, concept networks can be merged into a larger concept network that can be used to infer a user's likely intent given a query. In another embodiment, suggestions for related searches obtained using different concept networks can be merged or aggregated. Other users for concept networks in query processing, including assisted search, are also described.

Systems and methods for processing search requests include analyzing received queries in order to provide a more sophisticated understanding of the information being sought. In one embodiment, queries are parsed into units, which may comprise one or more words or tokens of the query, and the units are related in concept networks. Trend analysis is performed by sorting the queries into subsets along a dimension of interest and comparing concept networks for different subsets. Trend information is usable to enhance a response of an automated search agent to a subsequently received query.

Systems and methods for processing search requests include analyzing received queries in order to provide a more sophisticated understanding of the information being sought. In one embodiment, queries are parsed into units, which may comprise one or more words or tokens of the query, and the units are related in concept networks. Trend analysis is performed by sorting the queries into subsets along a dimension of interest and comparing concept networks for different subsets. Trend information is usable to enhance a response of an automated search agent to a subsequently received query.

Search results are processed using search requests, including analyzing received queries in order to provide a more sophisticated understanding of the information being sought. A concept network is generated from a set of queries by parsing the queries into units and defining various relationships between the units. From these concept networks, queries can be automatically categorized into categories, or more generally, can be associated with one or more nodes of a taxonomy. The categorization can be used to alter the search results or the presentation of the results to the user. As an example of alterations of search results or presentation, the presentation might include a list of “suggestions” for related search query terms. As other examples, the corpus searched might vary depending on the category or the ordering or selection of the results to present to the user might vary depending on the category. Categorization might be done using a learned set of query-node pairs where a pair maps a particular query to a particular node in the taxonomy. The learned set might be initialized from a manual indication of which queries go with which nodes and enhanced has more searches are performed. One method of enhancement involves tracking post-query click activity to identify how a category estimate of a query might have varied from an actual category for the query as evidenced by the category of the post-query click activity, e.g., a particular hits of the search results that the user selected following the query. Another method involved determining relationships between units in the form of clusters and using clustering to modify the query-node pairs.

Information retrieval systems face challenging problems with delivering highly relevant and highly inclusive search results in response to a user's query. Contextual personalized information retrieval uses a set of integrated methodologies that can combine automatic concept extraction / matching from text, a powerful fuzzy search engine, and a collaborative user preference learning engine to provide accurate and personalized search results. The system can include constructing a search query to execute a search of a database parsing an input query from a user conducting the search of the database into sub-strings, and matching the sub-strings to concepts in a semantic concept network of a knowledge base. The system can further map the matched concepts to criteria and criteria values that specify a set of constraints on and scoring parameters for the matched concepts. Furthermore, the system can learn user preferences to construct one or more profiles for producing personalized search results.

Information retrieval systems face challenging problems with delivering highly relevant and highly inclusive search results in response to a user's query. Contextual personalized information retrieval uses a set of integrated methodologies that can combine automatic concept extraction / matching from text, a powerful fuzzy search engine, and a collaborative user preference learning engine to provide accurate and personalized search results. The system can include constructing a search query to execute a search of a database. The system can parse an input query from a user conducting the search of the database into sub-strings, and can match the sub-strings to concepts in a semantic concept network of a knowledge base. The system can further map the matched concepts to criteria and criteria values that specify a set of constraints on and scoring parameters for the matched concepts.

In a search processing system, a concept network is generated from a set of queries by parsing the queries into units and defining various relationships between the units based in part on patterns of units that appear together in queries. Units in the concept network that have some similar characteristic(s) are grouped into superunits. For each superunit, there is a corresponding signature that defines the similar characteristic of the group. A query is processed by identifying constituent units, determining the superunit membership of some or all of the constituent units, and using that information to formulate a response to the query.

In a computerized search system, a document corpus is searched according to a query and results of the search are presentable as an ordered search result set comprising one or more search hits, a search hit being a document of the document corpus or reference thereto. Based on content of the search result set, a number of suggested queries and / or parameter options are presented to a querier, thereby allowing the querier to select a suggested query and / or provide values for the parameters and those selections can be used to reorder the hits of the search result set. The particular suggested queries and / or parameters presented depend on the page elements of the search result set's documents and their categorization. The categorization might be done using concept networks.

Systems and methods for processing search requests include analyzing received queries in order to provide a more sophisticated understanding of the information being sought. In one embodiment, queries are parsed into units, which may comprise one or more words or tokens of the query, and the units are related in concept networks. Trend analysis is performed by sorting the queries into subsets along a dimension of interest and comparing concept networks for different subsets. Trend information is usable to enhance a response of an automated search agent to a subsequently received query.

Techniques are provided for reranking and increasing the relevance of the results Internet searches. A search query is initially parsed into individual units. Each unit corresponds to one or more words that represents a natural concept. A concept network is analyzed to locate concepts that are related to the units in the search query. Particular concepts are selected from the concept network. Independent Internet searches are performed for each of the selected concepts. The search results from these searches are then compared to units in the original search query and ranked according to their relevance to the original search query.

A concept network that can be generated in response to a user query. In one embodiment, the concept network is being used with a search tool. The search tool searches a plurality of data storage locations. Each data storage location is arranged with a node. Certain ones of the nodes are connected by at least one link. The concept network selects a portion of certain ones of the nodes based on the link, wherein the at least one link is used for content purposes.

Techniques are provided for reranking and increasing the relevance of the results Internet searches. A search query is initially parsed into individual units. Each unit corresponds to one or more words that represents a natural concept. A concept network is analyzed to locate concepts that are related to the units in the search query. Particular concepts are selected from the concept network. Independent Internet searches are performed for each of the selected concepts. The search results from these searches are then compared to units in the original search query and ranked according to their relevance to the original search query.

A concept network that can be generated in response to a user query. Various embodiments include analysis of structure information, for example, where such information is based at least in part on Universal Resource Locators (URLs) of Web sites or data storage locations. A concept network may be used with a search tool where the search tool searches a plurality of sites (e.g., Web sites, data storage locations, etc.). In such an example, each site location is arranged with a node. Certain ones of the nodes are connected by at least one link. The concept network selects a portion of certain ones of the nodes based on the link, wherein the at least one link is used for content purposes.

Multiple concept networks are generated from subsets of received queries. These concept networks can be used in various ways to enhance response to subsequent queries. In one embodiment, concept networks can be merged into a larger concept network that can be used to infer a user's likely intent given a query. In another embodiment, suggestions for related searches obtained using different concept networks can be merged or aggregated. Other users for concept networks in query processing, including assisted search, are also described.

The invention discloses service assembly system and method based on service quality optimization and semantic information integration, which are designed mainly for improving the reliability and stability of service assembly and service quality. The service assembly system comprises a Web service network, an ontology concept network database, an index generating server, a service assembly engine server and a service assembly result executing server, wherein the Web service network provides service and a corresponding input / output data type; the ontology concept network database provides an ontology concept; the index generating server is used for establishing a service-data type-ontology index between the input / output type of the service and the ontology concept and storing the index in the index generating server; and the service assembly engine server is used for receiving the input / output data type demanded by the client end, inquiring the data type matched with the input / output data type from the service-data type-ontology index and carrying out service assembly according to a service assembly algorithm to obtain a service assembly result. The invention integrates semantic information, has optimal service quality and can efficiently process large-scale service assembly.

A system and method for building a multi-concept network based on web usage data that collect keywords used in a search site utilized by a plurality of users and web page information and build the multi-concept network for the keywords are provided. The method includes (a) collecting the keywords input by the users for searches in the site and the information on web pages read according to keyword search results; (b) for each keyword, selecting read web pages for each user; (c) for each keyword, setting each selected web page as one node, grouping the web page nodes for each user, connecting the web page nodes in a row, and arranging the web page nodes around the keyword; and (d) obtaining a similarity between two groups of the web page nodes arranged around the keyword, and integrating the two groups to form one group connected in a row when the similarity is above a predetermined standard value.With the system and method, web page usage data for each user for a user's interest keyword is collected to build a web page connection network. Thus, a web page connection network based on information on a variety of tendencies can be provided.

The invention discloses a visual question answering problem solving method based on a complex network analysis method, which includes semantic concept network construction, nonrandom depth walk, imageand text feature fusion and classifier, Semantic Concept Networks (SCNs) are designed to mine co-occurrence patterns of concepts to enhance semantic representation, non-random depth walk realizes themapping of complex networks to low-dimensional features, On the basis of constructing image semantic concept network, depth walk algorithm is used to learn the potential relationship of nodes in semantic concept network, and the nodes in complex network are mapped to a low-dimensional feature vector, and polynomial logistic regression is used to fuse image and text features to solve the visual question answering problem. The invention excavates the concept symbiosis mode and the hierarchical structure of the cluster concept, effectively integrates the visual and semantic features of the image, as well as the natural language features, and provides a feasible way for solving the visual question answering problem.

Described is a system for understanding machine-learning decisions. In an unsupervised learning phase, the system extracts, from input data, concepts represented by a machine-learning (ML) model in an unsupervised manner by clustering patterns of activity of latent variables of the concepts, where the latent variables are hidden variables of the ML model. The extracted concepts are organized into a concept network by learning functional semantics among the extracted concepts. In an operational phase, a subnetwork of the concept network is generated. Nodes of the subnetwork are displayed as a set of visual images that are annotated by weights and labels, and the ML model per the weights and labels.

The invention provides an automatic answer summarizing method and system for a question answering system. The method comprises the following steps: obtaining a user query question and a candidate answer set returned by the question answering system; extracting a concept from the question and building mapping of the concept and the question to obtain a question-concept set; extracting the concept from a candidate answer sentence and building mapping of the concept and the sentence to obtain a sentence-concept set; taking a union set of the concept sets of all sentences to obtain an answer-concept set; carrying out concept expansion and concept reduction on the question-concept set to obtain a question-expand-concept set; taking the union set of the question-expand-concept set and the answer-concept set to obtain a hit-concept set; and carrying out sentence quality calculation on the concept included in the hit-concept set to obtain an answer summary. Weights of the sentences and the like are built by the relationship between the concepts in a concept network; the problem that the sentence can be irrelevant to the question is solved; and the sentence which is the most relevant to the question is selected out from the answer through an integer programming method to form the final answer summary.

Systems and methods for providing a searchable concept network are provided. One such system includes a concept network application, which is hosted at least partially on a server. The concept network application may include or access an indexer for indexing text corpus. The text corpus is analyzed to extract concepts, as well as significance, relevancy and relationship information between concepts. The concepts and associated significance, relevancy and relationship information are utilized to construct a concept network, which is stored in one or more indexes or databases accessible to the concept network application. One or more user computer devices are provided access to the application, which receives a query from the user, and in response, provides the user with a relevant portion of the concept network. User behavior or interaction within the concept network may be detected and utilized in providing relevant advertisement information.

The invention discloses a feedback model information retrieval method and system based on conceptual network semanteme and a medium, and the method comprises the steps: obtaining a pseudo-related document set through providing a plurality of query keywords; processing the pseudo-correlation document set according to a weight retrieval model and semantic query of a conceptual network to obtain a plurality of target extension candidate words; performing linear fusion on the plurality of query keywords and the plurality of target extension candidate words to generate a query keyword set; and retrieving the query keyword set to obtain a final retrieval result. According to the feedback model information retrieval method and system based on conceptual network semantics and the medium, the userquery efficiency and effect are greatly improved.

The invention discloses an information retrieval method and system and a medium, and the method comprises the steps: providing a plurality of query keywords to acquire a pseudo-related document set; processing the pseudo-correlation document set according to a weight retrieval model and semantic query of a conceptual network to obtain a plurality of target extension candidate words; and retrievingthe plurality of target extension candidate words to obtain a final retrieval result. According to the information retrieval method and system and the medium disclosed by the invention, the user query efficiency and effect are greatly improved.

The invention discloses a construction method and device of a user knowledge concept network and an evaluation method of user knowledge.The construction method of the user knowledge concept network comprises the steps that firstly, each text contained in a text set containing m independent texts is preprocessed, and then each vocabulary of corpus serves as a concept subject term; all sentences and vocabularies are traversed, vocabularies appearing together with the concept subject terms in the same sentence are included into vocabulary sets corresponding to the concept subject terms, then vocabulary element screening is conducted on each vocabulary set, and a concept library is constructed; the field division is performed on concepts contained in the concept library by adopting a hierarchical clustering method; then, according to the matching condition of vocabularies contained in the user text data and a concept library, concepts contained in the user text data are obtained; and finally, a user knowledge concept network is constructed according to the concepts contained in the user text data and the divided concept fields. According to the method, the accuracy and objectivity of evaluation can be improved.

A method including steps of receiving a first concept network having a plurality of connected nodes; receiving a second concept network having a plurality of connected nodes; and generating a third concept network, the third concept network having a plurality of connected nodes being a combination of the first and second concept networks. Related systems and computer program products are also included.

The invention discloses a visual question-answering problem-solving method based on a complex network analysis method, including semantic concept network construction, non-random depth walk, image and text feature fusion, and a classifier. The semantic concept network construction aims to mine the co-occurrence of concepts Mode to enhance semantic expression, non-random deep walk realizes the mapping of complex network relations to low-dimensional features, on the basis of constructing image semantic concept network, apply deep walk algorithm to learn the potential relationship of nodes in semantic concept network, and combine complex The nodes in the network are mapped into a low-dimensional feature vector, and multinomial logistic regression fuses image and text features to solve visual question answering problems. The invention digs deep into the conceptual co-occurrence mode and the hierarchical structure of the cluster concept, effectively integrates the visual and semantic features of the image, and the natural language features, and provides a feasible way to solve the visual question answering problem.

Described is a system for understanding machine-learning decisions. In an unsupervised learning phase, the system extracts, from input data, concepts represented by a machine-learning (ML) model in an unsupervised manner by clustering patterns of activity of latent variables of the concepts, where the latent variables are hidden variables of the ML model. The extracted concepts are organized into a concept network by learning functional semantics among the extracted concepts. In an operational phase, a subnetwork of the concept network is generated. Nodes of the subnetwork are displayed as a set of visual images that are annotated by weights and labels, and the ML model per the weights and labels.