A search method using an artificial intelligence based information retrieval model and a method for training the artificial intelligence based information retrieval model used for the method are provided.

In the method, even if there is no labeled data and only a corpus exists, the artificial intelligence based information retrieval model can be trained using the weak-supervision methodology. Search can be performed by dividing documents into passages having short lengths. Compared to an information retrieval model based on unsupervised learning, improved search results are provided.

A method and system is disclosed for searching and displaying a learning chain including topics associated with a user. The method includes at an electronic device with a display: displaying a first user interface having one or more search fields for receiving a search query; receiving, at the one or more search fields, a first search query, wherein the first search query identifies a first topic; in response to receipt of the first search query, search a database. The database stores data identifying: a plurality of topics, including the first topic, a second topic different to the first topic and a third topic different to the first and second topics; a plurality of users, including a first user and a second user different to the first user; an association between the first topic and the first user; a plurality of learning chains of topics, including a first learning chain that includes the first topic and the second topic and not the third topic and wherein the learning chain indicates that the second topic is associated with the first topic as adjacent topics in the first learning chain; an association between the first learning chain and the first user. The method further includes displaying search results, the search results including the first user responsive to the stored association of the first topic to the first user; receiving a selection of the first user; in response to receiving selection of the first user, displaying the first learning chain, wherein the first learning chain includes a graphical representation of the first topic, a graphical representation of the second topic and a graphical representation of the association between the first topic and the second topic. A method and system is also disclosed for searching and displaying a learning chain associated with a course offered by a user and for generating a learning chain.

A method includes searching a technical document including a first data field, a second data field and a third data field based on search terms and search year ranges related to a technical field, generating a keyword set using the first data field, the second data field and the third data field of the searched technical document, scoring a plurality of keywords included in the keyword set, and selecting some of the plurality of keywords, re-searching the technical document related to the technical field, using the selected keywords, scoring the re-searched technical document to derive a representative document representing the technical field, and deriving a representative keyword representing the technical field, using the second data field included in the representative document, wherein the first data field includes a title of the technical document, the second data field includes a summary of the technical document, and the third data field includes keywords of the technical document.

In one embodiment, a method includes receiving, from a client system, a text input comprising one or more n-grams, determining, using a deep-learning model, a vector representation of the text input based on the one or more n-grams, determining an embedding of the vector representation of the text input in a d-dimensional embedding space, identifying one or more labels based on, for each of the one or more labels, a respective similarity of an embedding of a vector representation of the label in the embedding space to the embedding of the vector representation of the text input, and sending, to the client system in response to the received text input, instructions for presenting a user interface comprising one or more of the identified labels as suggested labels.

Methods and system for item matching are described. In one embodiment, compatibility-based text for an item may be accessed. A compatibility identifier may be identified based on the compatibility-based text. The compatibility identifier may be associated with an item cluster. The compatibility identifier may be used to identify a plurality of matching items. A result may be provided based on identification of the plurality of matching items. Additional methods and systems are disclosed.

A computer extracts from contact records that each include a contact identifier, a group identifier for each group with which the contact has had an interaction, and interaction information that indicates a number of interactions and a timing of a most recent interaction. The contact data records are processed to generate a contact profile record for each contact including group metric values and a corresponding value for each group metric value based on an interaction history of groups the contact has interacted with. An interaction analytics databases stores a set of contact profile records and group profile records for groups that include metric values associated with the group and an interaction history. They are processed with at least thousands of the contact profile records to determine group-contact compatibility factors. A compatibility parameter is generated and communicated for each of at least thousands of contacts based on the group-contact compatibility parameters.

An entity search method, a related device, and a computer storage medium are provided. The method includes: determining a first classifier and a second classifier that are included in query information; determining, based on a first entity library and the first classifier and the second classifier that are included in the query information, s correlations corresponding to each of w candidate entities, wherein information about the candidate entity includes a third classifier and a fourth classifier, and the correlation indicates a correlation between the classifiers in the query information and the classifiers in the candidate entity; and determining, based on the s correlations corresponding to each of the w candidate entities, information about a target entity corresponding to the query information, wherein the target entity is an entity in the w candidate entities.

Aspects of the present disclosure relate to interactive search training. A training canvas comprises results associated with a search query. The training canvas may be used as part of a training session that occurs during normal use of a search platform. When the search platform is first used, the results may be provided based on an existing model. An irrelevant result may be removed from the training canvas, such that a replacement result is added in its place. Additionally, results may be reordered, thereby indicating a ranking with which results should be displayed. Such interactions with the training canvas may be used to generate training data, such that a new model is trained accordingly. Thus, interactions with the training canvas yield high-quality training data that is usable to generate a model having equal or greater performance than a model that was trained using an equivalent amount of implicit training data.

According to one embodiment, a dialogue apparatus includes a processing circuit. The processing circuit designates one database from among a plurality of databases. The processing circuit extracts a keyword from text information. The processing circuit searches the designated database and another database which is included in the plurality of databases and which is other than the designated database, by using the keyword. The processing circuit generates a response in accordance with a first number of hit items, which is a number of data items matching the keyword in the designated database, and a second number of hit items, which is a number of data items matching the keyword in said another database.

Embodiments herein describe creating a searchable catalog of composite actions for embroidery designs. The catalog can also store a design intent and the embroidery technique for each of the composite actions. The design intents and the embroidery techniques for the composite actions can be stored in a design intent hierarchy and an embroidery technique hierarchy which can then be compared to text in a search query to identify matches. These matches can then be used to identify composite actions that match the search query.