Ranking search queries using contextual relevance and third-party factors
A computer system allowing users to search for items of interest provides a search query interface. The system receives characters of a search query in the search interface as the user enters the characters and interactively calculates, ranks, and displays a set of possible search query options from which the user can select. To rank the set of possible search query options, the system modifies rankings of candidate search queries based on factors associated with third parties. More specifically, contextual relevance scores are computed for the candidate search queries based on the context, such as a user to whom the search results are provided. These contextual relevance scores are in turn adjusted using factors associated with third parties, such as values calculated based on consideration offered by third parties. Users are shown the search query options, ranked in order of the adjusted relevance scores, as possible query selections.
1 . A method of providing search results, performed at a computer system comprising a processor and a computer-readable medium storing a database of historical search queries provided by one or more users of an online system, the method comprising:
causing presentation of a search interface presented on a client device of a user, wherein the search interface includes an input for inputting a search query;
receiving a partial search query entered into the input of the search interface presented on the client device;
identifying, from the database of historical search queries, a plurality of candidate search queries matching the partial search query, wherein each candidate search query includes all text of the partial search query and additional text not in the partial search query;
generating, for each of the candidate search queries, a contextual relevance score based on whether a user has searched, viewed, or purchased one or more items associated with the candidate search queries;
generating, for each of the candidate search queries, an estimated value of consideration offered by third parties based on consideration obtained by third parties in response to user selection of historical search queries;
adjusting, for each of the candidate search queries and using at least the estimated values of consideration, the contextual relevance scores;
generating a listing of at least a subset of the candidate search queries, the subset of the candidate search queries selected and sorted based on the adjusted contextual relevance scores;
causing the client device of the user to display an updated search interface presenting the generated listing adjacent to the input partial search query;
receiving user input, via the search interface, selecting one of the candidate search queries from the listing;
in response to the user input selecting the one candidate search query from the listing, causing presentation, via the search interface presented on the client device, content items associated with the selected candidate search query;
receiving user input, via the search interface, adding one or more of the content items presented in the search interface to an order for the user, the order identifying one or more content items to be obtained at a retailer location; and
transmitting the order identifying the one or more content items to be obtained at the retailer location to another client device associated with a picker, wherein the picker performs collection of the content items at the retailer location.
2 . The method of claim 1 , further comprising:
generating expected gross merchandise values associated with the candidate search queries; and
adjusting, at least in part using the expected gross merchandise values, the contextual relevance scores.
3 . The method of claim 1 , further comprising:
determining unused amounts of consideration budgets of the third parties for search queries; and
generating, at least in part based on the determined unused amounts, the listing of the candidate search queries.
4 . The method of claim 1 , further comprising:
reducing the listing of the candidate search queries responsive to the candidate search queries containing brand names.
5 . The method of claim 1 , wherein generating the estimated values comprises applying a trained model to identify the estimated values, the trained model taking as input at least one of: user identifier of the user, day, time, or the partial search query.
6 . The method of claim 1 , wherein the estimated values of consideration are generated with a machine-learned model, the method further comprising retraining the machine-learned model based on a user selection of one of the plurality of candidate search queries.
7 . The method of claim 1 , further comprising:
applying, for each candidate search query, an item availability model to predict a likelihood of item availability for each content item associated with the candidate search query; and
adjusting further the contextual relevance score for each candidate search query based on the predicted likelihoods of item availability for the one or more content items associated with the candidate search query.
8 . A non-transitory computer-readable storage medium storing a database of historical search queries provided by one or more users of an online system and instructions that, when executed by a computer processor, cause the computer processor to perform actions comprising:
causing presentation of a search interface presented on a client device of a user, wherein the search interface includes an input for inputting a search query;
receiving a partial search query entered into the input of the search interface presented on the client device;
identifying, from the database of historical search queries, a plurality of candidate search queries matching the partial search query, wherein each candidate search query includes all text of the partial search query and additional text not in the partial search query;
generating, for each of the candidate search queries, a contextual relevance score based on whether a user has searched, viewed, or purchased one or more items associated with the candidate search queries;
generating, for each of the candidate search queries, an estimated value of consideration offered by third parties based on consideration obtained by third parties in response to user selection of historical search queries;
adjusting, for each of the candidate search queries and using at least the estimated values of consideration, the contextual relevance scores;
generating a listing of at least a subset of the candidate search queries, the subset of the candidate search queries selected and sorted based on the adjusted contextual relevance scores;
causing the client device of the user to display an updated search interface presenting the generated listing adjacent to the input partial search query;
receiving user input, via the search interface, selecting one of the candidate search queries from the listing;
in response to the user input selecting the one candidate search query from the listing, causing presentation, via the search interface presented on the client device, content items associated with the selected candidate search query;
receiving user input, via the search interface, adding one or more of the content items presented in the search interface to an order for the user, the order identifying one or more content items to be obtained at a retailer location; and
transmitting the order identifying the one or more content items to be obtained at the retailer location to another client device associated with a picker, wherein the picker performs collection of the content items at the retailer location.
9 . The non-transitory computer-readable storage medium of claim 8 , the actions further comprising:
generating expected gross merchandise values associated with the candidate search queries; and
adjusting, at least in part using the expected gross merchandise values, the contextual relevance scores.
10 . The non-transitory computer-readable storage medium of claim 8 , the actions further comprising:
determining unused amounts of consideration budgets of the third parties for search queries; and
generating, at least in part based on the determined unused amounts, the listing of the candidate search queries.
11 . The non-transitory computer-readable storage medium of claim 8 , the actions further comprising:
reducing the listing of the candidate search queries responsive to the candidate search queries containing brand names.
12 . The non-transitory computer-readable storage medium of claim 8 , wherein generating the estimated values comprises applying a trained model to identify the estimated values, the trained model taking as input at least one of: user identifier of the user, day, time, or the partial search query.
13 . A computer system comprising:
a computer processor; and
a non-transitory computer-readable storage medium storing instructions that when executed by the computer processor perform actions comprising:
causing presentation of a search interface presented on a client device of a user, wherein the search interface includes an input for inputting a search query;
receiving a partial search query entered into the input of the search interface presented on the client device of;
identifying, from a database of historical search queries, a plurality of candidate search queries matching the partial search query, wherein each candidate search query includes all text of the partial search query and additional text not in the partial search query;
generating, for each of the candidate search queries, a contextual relevance score based on whether a user has searched, viewed, or purchased one or more items associated with the candidate search queries;
generating, for each of the candidate search queries, an estimated value of consideration offered by third parties based on consideration obtained by third parties in response to user selection of historical search queries;
adjusting, for each of the candidate search queries and using at least the estimated values of consideration, the contextual relevance scores;
generating a listing of at least a subset of the candidate search queries, the subset of the candidate search queries selected and sorted based on the adjusted contextual relevance scores;
causing the client device of the user to display an updated search interface presenting the generated listing adjacent to the input partial search query;
receiving user input, via the search interface, selecting one of the candidate search queries from the listing;
in response to the user input selecting the one candidate search query from the listing, causing presentation, via the search interface presented on the client device, content items associated with the selected candidate search query;
receiving user input, via the search interface, adding one or more of the content items presented in the search interface to an order for the user, the order identifying one or more content items to be obtained at a retailer location; and
transmitting the order identifying the one or more content items to be obtained at the retailer location to another client device associated with a picker, wherein the picker performs collection of the content items at the retailer location.
14 . The computer system of claim 13 , the actions further comprising:
generating expected gross merchandise values associated with the candidate search queries; and
adjusting, at least in part using the expected gross merchandise values, the contextual relevance scores.
15 . The computer system of claim 13 , the actions further comprising:
determining unused amounts of consideration budgets of the third parties for search queries; and
generating, at least in part based on the determined unused amounts, the listing of the candidate search queries.
16 . The computer system of claim 13 , the actions further comprising:
reducing the listing of the candidate search queries responsive to the candidate search queries containing brand names.
17 . The computer system of claim 13 , wherein generating the estimated values comprises applying a trained model to identify the estimated values, the trained model taking as input at least one of: user identifier of the user, day, time, or the partial search query.
18 . The computer system of claim 13 , wherein the estimated values of consideration are generated with a machine-learned model, the actions further comprising retraining the machine-learned model based on a user selection of one of the plurality of candidate search queries.