Abstract: The present invention extends to methods, systems, and computer program products for training a classification model to predict categories. In one implementation, a method identifies category mappings generated for dominant queries associated with a query log. The method identifies mappings between a first set of queries and categories shown for the first set of queries, and identifies mappings between a second set of queries and clicked products for the second set of queries. A classification model is trained based on the mappings generated for dominant queries, the mappings between queries and the shown categories, and the mappings between queries and the clicked products.

Abstract: The present invention extends to methods, systems, and computer program products for training a classification model to predict categories. In one implementation, a method identifies category mappings generated for dominant queries associated with a query log. The method identifies mappings between a first set of queries and categories shown for the first set of queries, and identifies mappings between a second set of queries and clicked products for the second set of queries. A classification model is trained based on the mappings generated for dominant queries, the mappings between queries and the shown categories, and the mappings between queries and the clicked products.

Abstract: Queries are analyzed to identify start-side sub-queries that include a starting word of the query and one or more contiguous words and end-side sub-queries that include an ending word of the query and one or more contiguous words. A start-side count is determined for a sub-query indicating a number of times it occurs among the queries as a start-side query. An end-side count is also computed for each sub-query indicating the number of times it occurs as an end-side sub-query. A score is computed for each sub-query that decreases with the start-side count and increases with the end-side count. A search is performed for a received query with some of the phrases (e.g. noun phrases) thereof weighted according to the scores. Noun phrases may be identified as the longest n-gram for which an entry in a reference dictionary exists.

Abstract: Ranking search results based on color including evaluating whether a color term in a search query was intended to identify an actual color and returning different search results based on that determination. When an actual color was intended, an initial product ranking is updated based on color similarity of the product. In some embodiments, a color similarity score is calculated using a color difference formula, a cosine similarity measurement is determined of a cosine of an angle between a query vector and a title vector, and the color is considered to have been intended to identify the actual color when the cosine similarity measurement is less than a threshold value. In particular embodiments, the query vector includes individual terms of the search query and a number of occurrences of each individual term. Different embodiments include computer implemented systems and methods.

Abstract: The present invention extends to methods, systems, and computer program products for training a classification model to predict categories. In one implementation, a method identifies category mappings generated for dominant queries associated with a query log. The method identifies mappings between a first set of queries and categories shown for the first set of queries, and identifies mappings between a second set of queries and clicked products for the second set of queries. A classification model is trained based on the mappings generated for dominant queries, the mappings between queries and the shown categories, and the mappings between queries and the clicked products.

Abstract: The present invention extends to methods, systems, and computer program products for training a classification model to predict categories. In one implementation, a method identifies category mappings generated for dominant queries associated with a query log. The method identifies mappings between a first set of queries and categories shown for the first set of queries, and identifies mappings between a second set of queries and clicked products for the second set of queries. A classification model is trained based on the mappings generated for dominant queries, the mappings between queries and the shown categories, and the mappings between queries and the clicked products.

Abstract: The present invention extends to methods, systems, and computer program products for training a classification model to predict categories. In one implementation, a method identifies category mappings generated for dominant queries associated with a query log. The method identifies mappings between a first set of queries and categories shown for the first set of queries, and identifies mappings between a second set of queries and clicked products for the second set of queries. A classification model is trained based on the mappings generated for dominant queries, the mappings between queries and the shown categories, and the mappings between queries and the clicked products.

Abstract: The present invention extends to methods, systems, and computer program products for improving search relevance. In one implementation, a method identifies a first group of mappings generated for dominant queries associated with a query log. The method also identifies a second group of mappings that includes mappings between a first set of queries and categories shown for the first set of queries, and mappings between a second set of queries and clicked products for the second set of queries. Upon receipt of a query, the method applies the second group of mappings for the received query if the category mappings for the received query are in the second group of mappings. Alternatively, the method applies the first group of mappings for the received query if the category mappings for the received query are in the first group of mappings but not in the second group of mappings.

Abstract: Example systems and methods that rank search results based on color are described. In one implementation, a method receives a product search query from a user and identifies a reference to a color in the product search query. The product search query is executed to identify multiple products. The method further identifies references to the color in a title or a description of each of the multiple products. A ranking of the multiple products is created by increasing the ranking of products that include a reference to the color in the title or the description.

Abstract: Example systems and methods that rank search results based on color are described. In one implementation, a method receives a product search query from a user and identifies a reference to a color in the product search query. The product search query is executed to identify multiple products. The method further identifies references to the color in a title or a description of each of the multiple products. A ranking of the multiple products is created by increasing the ranking of products that include a reference to the color in the title or the description.

Abstract: Ranking search results based on color including evaluating whether a color term in a search query was intended to identify an actual color and returning different search results based on that determination. When an actual color was intended, an initial product ranking is updated based on color similarity of the product. In some embodiments, a color similarity score is calculated using a color difference formula, a cosine similarity measurement is determined of a cosine of an angle between a query vector and a title vector, and the color is considered to have been intended to identify the actual color when the cosine similarity measurement is less than a threshold value. In particular embodiments, the query vector includes individual terms of the search query and a number of occurrences of each individual term. Different embodiments include computer implemented systems and methods.

Abstract: Example systems and methods that select search result images are described. In one implementation, a method accesses a ranking of multiple products associated with a search query and identifies a reference to a color in the search query. The method identifies a first product from the ranking of multiple products and identifies multiple product images associated with the first product. A color determination is made regarding each of the multiple product images. A product image having an associated color that is most similar to the color in the search query is selected for presentation in the search results.

Abstract: Queries are analyzed to identify start-side sub-queries that include a starting word of the query and one or more contiguous words and end-side sub-queries that include an ending word of the query and one or more contiguous words. A start-side count is determined for a sub-query indicating a number of times it occurs among the queries as a start-side query. An end-side count is also computed for each sub-query indicating the number of times it occurs as an end-side sub-query. A score is computed for each sub-query that decreases with the start-side count and increases with the end-side count. A search is performed for a received query with some of the phrases (e.g. noun phrases) thereof weighted according to the scores. Noun phrases may be identified as the longest n-gram for which an entry in a reference dictionary exists.

Abstract: A method for merging two taxonomies is disclosed. Top levels of first and second taxonomies are merged. For the second taxonomy nodes are evaluated and selectively merged with nodes of the first taxonomy according to comparison of scores for these nodes with a threshold. The score for a node of the first taxonomy is a combination of one or more of a lineage quality score, Jaccard distance, string edit distance, and category depth score. After an iteration, mergings between nodes of the first and second taxonomies may be reversed if child nodes of the merged nodes were not likewise merged. Iterations may be repeated until no nodes are merged in an iteration.

Abstract: The present invention extends to methods, systems, and computer program products for classifying e-commerce queries to generate category mappings for dominant products. A query log is mined for any query records with click through information that indicates one or more products were selected from among corresponding query results. For each of one or more categories, a selection rate is calculated for any product selected from among at least one corresponding query result returned in response to a query of the category. A specified top number of products are identified in the category. The specified top number of products has higher selection rates relative to other products in the category. A category score is calculated for the category based on product information associated with the specified top number of products in the category. The one or more categories are ranked based on the calculated category scores.

Abstract: The present invention extends to methods, systems, and computer program products for training a classification model to predict categories. In one implementation, a method identifies category mappings generated for dominant queries associated with a query log. The method identifies mappings between a first set of queries and categories shown for the first set of queries, and identifies mappings between a second set of queries and clicked products for the second set of queries. A classification model is trained based on the mappings generated for dominant queries, the mappings between queries and the shown categories, and the mappings between queries and the clicked products.

Abstract: Example systems and methods that classify queries to generate category mappings are described. In one implementation, a method mines a query log to identify query records with click through information and display information that indicate one or more products were selected or displayed within a specified date range. A category is associated with each of the products. The method calculates a display rate and a selection rate for any product selected or displayed among the products. A mapping is identified between queries and the shown categories, and between queries and the clicked products. A category score is calculated for a particular category based on a number of times the category is shown and a number of times the category is clicked.

Abstract: The present invention extends to methods, systems, and computer program products for improving search relevance. In one implementation, a method identifies a first group of mappings generated for dominant queries associated with a query log. The method also identifies a second group of mappings that includes mappings between a first set of queries and categories shown for the first set of queries, and mappings between a second set of queries and clicked products for the second set of queries. Upon receipt of a query, the method applies the second group of mappings for the received query if the category mappings for the received query are in the second group of mappings. Alternatively, the method applies the first group of mappings for the received query if the category mappings for the received query are in the first group of mappings but not in the second group of mappings.

Abstract: Example systems and methods that rank search results based on color are described. In one implementation, a method receives a product search query from a user and identifies a reference to a color in the product search query. The product search query is executed to identify multiple products. The method further identifies references to the color in a title or a description of each of the multiple products. A ranking of the multiple products is created by increasing the ranking of products that include a reference to the color in the title or the description.

Abstract: Example systems and methods that rank search results based on color are described. In one implementation, a method receives a product search query from a user and identifies a reference to a color in the product search query. The product search query is executed to identify multiple products. The method further identifies references to the color in a title or a description of each of the multiple products. A ranking of the multiple products is created by increasing the ranking of products that include a reference to the color in the title or the description.