Abstract: Methods and systems for enabling access to listing performance data, used in conjunction with a search engine to order search results, are described. Consistent with some embodiments, a listing performance data engine module processes API commands received from client computing systems. The data engine module processes the API commands to identify and communicate relevant listing performance data that is used by a listing performance algorithm to assign a listing performance score to item listings satisfying a search request. The listing performance data provides sellers who have items listed for sale an insight as to how their item listings are performing in comparison to other item listings.

Abstract: A system and method, which may be an offline method, extracts relevant image features about listing items in a network-based publication system for enabling image similarity searching of such listing items. When a seller lists an item, an image of the item is uploaded and may be sent to a picture processing service, which generates several digests. The digests are compressed data structures each representing a particular image feature such as edge, color, texture, or words. These digests are then stored in a search database, where the digests can be used to retrieve listings by image similarity at scale. A similar process can be performed for a query listing for searching the search database for items similar to the query listing.

Abstract: A method and system for determining an order of presentation of the search results is provided. An example system comprises a search request detector, a search engine, and an intermingler. The search request detector may be configured to receive a search request, and the search engine may be configured to determine search results based on the search request. The search results may comprise a first set of items of a first type and a second set of items of a second type. The number of items in the first set and the number of items in the second may reflect a predetermined target ratio. The intermingler may be configured to determine an order of presentation of the search results in a list of search results based on the target ratio.

Abstract: Methods and systems for analyzing, ordering and presenting item listings are described. In one example embodiment, a search query is processed to identify item listings satisfying the search query. Then, for each item listing that satisfies the search query, a ranking score is derived and assigned to the item listing. The ranking score is based in part on a relevance score, a listing quality score and a business rules score (or, adjustment factor). Finally, the item listings are ordered, based on their corresponding ranking score, and presented in order in a search results page.

Abstract: A system and method, which may be an offline method, extracts relevant image features about listing items in a network-based publication system for enabling image similarity searching of such listing items. When a seller lists an item, an image of the item is uploaded and may be sent to a picture processing service, which generates several digests. The digests are compressed data structures each representing a particular image feature such as edge, color, texture, or words. These digests are then stored in a search database, where the digests can be used to retrieve listings by image similarity at scale. A similar process can be performed for a query listing for searching the search database for items similar to the query listing.

Abstract: Methods and systems for analyzing, ordering and presenting item listings are described. In one example embodiment, a search query is processed to identify item listings satisfying the search query. Then, for each item listing that satisfies the search query, a ranking score is derived and assigned to the item listing. The ranking score is based in part on a relevance score, a listing quality score and a business rules score (or, adjustment factor). Finally, the item listings are ordered, based on their corresponding ranking score, and presented in order in a search results page.

Abstract: A system and method, which may be an offline method, extracts relevant image features about listing items in a network-based publication system for enabling image similarity searching of such listing items. When a seller lists an item, an image of the item is uploaded and may be sent to a picture processing service, which generates several digests. The digests are compressed data structures each representing a particular image feature such as edge, color, texture, or words. These digests are then stored in a search database, where the digests can be used to retrieve listings by image similarity at scale. A similar process can be performed for a query listing for searching the search database for items similar to the query listing.

Abstract: Methods and systems for enhancing a user interface of a client device are described. In an example embodiment, a machine processes a query to identify one or more item listings that satisfy the query. Each item listing is associated with at least one item being offered for sale. The machine assigns one or more scores to one or more item listings that satisfy the query based on an observed demand metric derived from historical performance data of the one or more item listings. The machine causes a presentation of the one or more item listings that satisfy the query in a search results page displayed in the user interface of the client device. The one or more item listings are positioned in the search results page based on the one or more scores assigned to the one or more item listings.

Abstract: A system and method extracts relevant image features about listed products in a network-based publication system for enabling image similarity searching of such listed products. When a seller lists a product, an image of the product is uploaded and may be sent to a picture processing service, which generates several digests. The digests are compressed data structures each representing a particular image feature of a listed product such as edge, color, texture, or words. These digests are then stored in a database, where the digests can be used to retrieve listings of products by image similarity at scale. A similar process can be performed for a query by image similarity searching the database for products similar to the query. When a product is located by image similarity search, a selectable icon is provided to enable more products like the located product to be located by image similarity search.

Abstract: Methods and systems for assessing the quality of an item listing are described. In an example embodiment, a listing quality score for an item listing is derived as a weighted sum of first and second parts. The first part represents a predicted score based on a comparison of item attributes for the item listing that are known at listing time, with item attributes of similar item listings that have historical data available for assessing their actual performance. The second part is based on one or more observed demand metrics representing actual historical performance of the item listing. The weighting factor is derived, such that over time, the emphasis shifts from the predicted to the observed score.

Abstract: Using a processor, receiving, a query including a query image. A database is searched for a set of images similar to the query image, using a two-pass search. The results of the searching are then provided, the results including image members of the set of images similar to the query image. The first pass may be performed using a TF-IDF algorithm and the second pass ranks a predetermined number of results from the first pass by a best match algorithm or other type of algorithm. The type of second pass algorithm may be selectable by a user. If an end signal is not detected, a further two-pass search may be made based on a subsequent query having a subsequent query image. The subsequent query image may be one of the results of the previous two pass search. A subsequent query image may change from an image of a product of a first product category to an image of a product of a second product category.

Abstract: A method, a system and a computer program product for ranking reviewable entities based on sentiment expressed about the entities. A plurality of review texts are identified wherein each review text references an entity. A plurality of sentiment scores associated with the plurality of review texts are generated, wherein each sentiment score for a review text indicates a sentiment directed to the entity referenced by the review text. A plurality of ranking scores for the plurality of entities are generated wherein each ranking score is based at least in part on one or more sentiment scores associated with one or more review texts referencing the entity. A plurality of search results associated with the plurality of entities are displayed based at least in part on the ranking scores.

Abstract: A method and system are provided for a part-of-speech tagger that may be particularly useful for resource-poor languages. Use of manually constructed tag dictionaries from dictionaries via bitext can be used as type constraints to overcome the scarcity of annotated data in some instances. Additional token constraints can be projected from a resource-rich source language via word-aligned bitext. Several example models are provided to demonstrate this such as a partially observed conditional random field model, where coupled token and type constraints may provide a partial signal for training. The disclosed method achieves a significant relative error reduction over the prior state of the art.

Abstract: A source language sentence is tagged with non-lexical tags, such as part-of-speech tags and is parsed using a lexicalized parser trained in the source language. A target language sentence that is a translation of the source language sentence is tagged with non-lexical labels (e.g., part-of speech tags) and is parsed using a delexicalized parser that has been trained in the source language to produce k-best parses. The best parse is selected based on the parse's alignment with lexicalized parse of the source language sentence. The selected best parse can be used to update the parameter vector of a lexicalized parser for the target language.

Abstract: A system and method, which may be an offline method, extracts relevant image features about images in a network-based publication system for enabling image similarity searching of such images. An image is uploaded and may be sent to a picture processing service, which generates digests. The digests are compressed data structures each representing a particular image feature such as edge, color, texture, or words. These digests are then stored in a search database, where the digests can be used to retrieve images by image similarity at scale. A similar process can be performed for an image query for searching the search database for images similar to the image.

Abstract: Methods and systems for using a demand metric for disambiguating a keyword entered from a user of a client machine, the keyword ambiguously identifying a product category, deriving a dominant product category that indicates the probable identity of the product category, and providing selectable images of aspects of the dominant product category for rendering at the client machine. Selection of one of the selectable images of aspects of the dominant product category disambiguates the identity of the ambiguously identified product category. Additional aspects of the disambiguated product category are then presented to the user.

Abstract: Methods and systems for using a demand metric for disambiguating a keyword entered from a client machine, the keyword identifying a product brand and ambiguously identifying a product category of the product brand; computing the dominant product category that indicates the probable identity of the product category; and providing signals representing selectable images of aspects of the dominant product category for rendering at the client machine.

Abstract: Methods and systems for analyzing, ordering and presenting item listings are described. In one example embodiment, a search query is processed to identify item listings satisfying the search query. Then, for each item listing that satisfies the search query, a ranking score is derived and assigned to the item listing. The ranking score is based in part on a relevance score, a listing quality score and a business rules score (or, adjustment factor). Finally, the item listings are ordered, based on their corresponding ranking score, and presented in order in a search results page.

Abstract: A weighted combination of attributes including but not limited to color, shape, pattern, brand, style, size, and material may be used to rank items when searching for similar items of a given category on an ecommerce system website. The weights on each attribute may be optimized by using feedback from users, the feedback indicating the importance of each attribute to a purchaser when searching for similar items of various categories. Alternatively, click-through patterns from the ecommerce system website can be mined and used to optimize the importance of each attribute to users by item category. The optimum attribute weights for that items category are used to rank the results in a search for similar items in that category.

Abstract: A source language sentence is tagged with non-lexical tags, such as part-of-speech tags and is parsed using a lexicalized parser trained in the source language. A target language sentence that is a translation of the source language sentence is tagged with non-lexical labels (e.g., part-of speech tags) and is parsed using a delexicalized parser that has been trained in the source language to produce k-best parses. The best parse is selected based on the parse's alignment with lexicalized parse of the source language sentence. The selected best parse can be used to update the parameter vector of a lexicalized parser for the target language.