Wal-mart Stores, Inc.

Jinxing Cheng

Abhishek Gattani

Systems and methods are disclosed herein for providing search results, particularly product records from a product database. Tokens of a query are identified. Product records including one or more tokens in values associated with the attributes thereof are identified. A score is generated for attributes of products including a token (an attribute-token score). The attribute-token score is based on a frequency of occurrence of the token in the attribute value and a length of the attribute value in which it occurs. The attribute-token scores for the token are summed to generated a score. The token score for a token may be scaled in a non-linear fashion such that the scaled value approaches a saturation value, which may be an inverse document frequency (IDF) score for the token. The scaled values may then be summed to generate a score for the document that is used to rank the search results.

Attribute-based document searching

Systems and methods are disclosed herein for providing search results, particularly product records from a product database. Past queries are analyzed and grouped into clusters according to similarities, with each query including a highly relevant head query and a plurality of tail queries. Similarity of queries may be determined based on user response similarity to query results, co-occurrence, and textual similarity. One or more categories are identified for the clusters, such as based on click-through rates for search results of the head queries of the clusters. Upon receiving a query, a cluster for the query is identified, such as according to similarity to one or more queries of the cluster. The categories associated with the cluster are then used to one or both of augment the query and rank search results for the query.

Search result ranking using query clustering

Systems and methods are disclosed herein for providing search results, particularly product records from a product database. Past queries are analyzed and grouped into clusters according to similarities with one another. For an individual query, relevant search results are identified and departments corresponding thereto are also identified. Confidence scores are determined for the identified departments and if one or more departments have confidence scores above a threshold, the search results are filtered based thereon. If not, then departments corresponding to a query cluster relevant to the query are selected. Additional departments may also be selected based on the number of search results corresponding to the additional departments. Search results filtered according to one or more selected departments are transmitted to an author of the query.

Search result ranking by department

Systems and methods are disclosed herein for providing search results, particularly product records from a product database. Products records have facet groups associated therewith and facet groups have probabilities associated therewith indicating the usefulness of the facet group as a way of organizing or filtering search results. For a query, relevant product records are identified. For facet groups present among the relevant product records entropy values are calculated based on a distribution of values for the attribute defining the facet group. The entropy value increases as the distribution of values becomes more uniform. The facet group probability and entropy value are combined to assign a score to the facet groups. The facet groups are ranked and selected according to the score and search results are presented with indicators of the facet groups to enable a user to manage the search results.

Growing community of inventors

Palo Alto, CA, United States of America

Jinxing Cheng