30 results about "Semantic matchmaking" patented technology

A method, apparatus and computer program product for presenting a user interface for a conversational system is described. A user input is received in a dialog between a user and the conversational system, the user input in a natural language. A domain trained semantic matcher is used to determine a set of entities and a user intent from the user input. One or more queries is generated to selected ones of a plurality of knowledge sources, the knowledge sources created from domain specific knowledge. The results from the one or more queries are ranked based on domain specific knowledge. A system response is presented in the dialog based on at least a highest ranked result from the plurality of knowledge sources.

Disclosed is an implementation method of machine reading comprehension, filtering the first data set according to the question type to obtain the filtered second data set; performing semantic matching between the question and the data i in the second data set to obtain the semantic matching score of the data i; Calculate the maximum coverage of the question and the data i to obtain the feature matching score of the data i; weight the semantic matching score and feature matching score of the data i to obtain the matching score of the question and the data i; calculate the data i according to the multi-document voting algorithm The voting score of data i, according to the matching score and voting score of data i, calculate the final score of data i; select the first n data in the sequence according to the descending order of the final score as the candidate data set; input the candidate data set into the baseline The model predicts the answer of the input data set based on the baseline model, and obtains a candidate answer set for the question. This application implements the sorting of data sets and extracts effective answers to questions.

The disclosure discloses a training method, device and electronic equipment for a semantic vector extraction model, which relate to the field of artificial intelligence, and in particular to the technical fields of deep learning and natural language processing. The specific implementation method is: obtain pre-training text, and based on the pre-training text, train the semantic expression extraction network to generate a pre-semantic expression extraction network; obtain sensitive text samples, and input the sensitive text samples into the pre-semantic Recognize in the expression recognition network to obtain the semantic vector of the sensitive text sample; input the semantic vector into the semantic matching network for training, and adjust the pre-semantic expression extraction network and the semantic expression based on the loss value of each training The matching network is used to generate a semantic vector extraction model, which avoids the over-fitting problem that the semantic vector extraction model is prone to when training with small sample data, and improves the efficiency and reliability of the semantic vector extraction model training process.

The invention relates to a deep learning-based question and answer matching method. The method comprises the following steps of: 1) sufficiently learning word orders and sentence local features of a question text and an answer text by utilizing two underlying deep neural networks: a long short-term memory network LSTM and a convolutional neural network CNN; and 2) selecting a keyword with best semantic matching on the basis of a pooling manner of an attention mechanism AM. Compared with existing methods, the method has the advantages of being in low in feature engineering workload, strong in cross-field performance and relatively high in correctness, and can be effectively applied to the fields of commercial intelligent customer service robots, automatic driving, internet medical treatment, online forum and community question answering.

The invention discloses a multi-platform data release method, comprising: a preset database, the database includes a plurality of first data fields; acquiring release files of each platform, and the release files include platform labels, release rules, multiple second data fields, The second data attribute corresponding to the second data field one-to-one, establishes a platform label, a publishing rule, and a plurality of second data field associations; performs semantic matching on the second data attribute and the second data field of a plurality of publishing files, and Judging whether the degree of matching reaches the threshold; if so, establishing associations between the second data fields between the platforms; performing semantic matching between the first data field and the second data field, and judging whether the degree of matching reaches the threshold; if so, establishing the second data field A data field is associated with a second data field; the published data is obtained, and the data is released according to the publishing rules and the association of the fields. It can realize the automatic matching of the data release file and the matching database, and the automatic release of the data file.

The invention discloses a voice information processing method and device, wherein the voice information processing method includes: acquiring the voice information of the target user; performing first semantic matching on the voice information based on the target user semantic library; when the first semantic matching When the result is that the matching is not successful, obtain the synonymous voice information of the target user corresponding to the voice information; perform the second semantic matching on the synonymous voice information based on the target user semantic database; when the result of the second semantic matching is that the matching is successful , acquire the target semantics matching the synonymous voice information; store the corresponding relationship between the voice information, the synonymous voice information and the target semantics in the target user semantic database. The present invention establishes a target user semantic library that matches the target user's speaking habits during the actual voice interaction process of the target user, thereby reducing the difficulty of interaction, improving the efficiency of interaction and the accuracy of interaction results, and improving user experience.

During operation, the system obtains a first embedding produced by an embedding model from an input string representing an entity and a hierarchy of clusters of embeddings generated by the embedding model from a set of standardized entities. Next, the system searches the hierarchy of clusters for a subset of the embeddings that are within a threshold proximity to the first embedding in a vector space. The system then calculates embedding match scores between the input string and a first subset of the standardized entities represented by the subset of the embeddings based on distances between the subset of the embeddings and the first embedding in the vector space. Finally, the system modifies, based on the embedding match scores, content outputted in response to the input string within a user interface of an online system.