Searchable index
Systems and techniques are disclosed for generating entries for a searchable index based on rules generated by one or more machine-learned models. The index entries can include one or more tokens correlated with an outcome and an outcome probability. A subset of tokens can be identified based on the characteristics of an event. The index may be searched for outcomes and their respective probabilities that correspond to tokens that are similar to or match the subset of tokens based on the event.
1. A computer-implemented method comprising:
before receiving, at a search engine system, a subsequent search query that is associated with at least one or more features:
generating, using a predictive model and by the search engine system, data indicating a predicted likelihood of a user selecting a particular item of content after submitting a search query that is associated with the one or more features, wherein the particular item of content is a result for the search query, wherein the predicted likelihood is generated by the predictive model based at least on the one or more features, and wherein the one or more features include at least one feature relating to the user, the search query submitted by the user, the particular item of content, a device associated with the user, or a current condition; and
storing, by the search engine system and in a record entry of a database for a search engine index, data that associates the particular item of content and the one or more features with the predicted likelihood generated by the predictive model of the user selecting the particular item of content after submitting the search query that is associated with the one or more features;
receiving, at the search engine system, the subsequent search query that is associated with at least the one or more features;
in response to the subsequent search query:
accessing, by the search engine system, the record entry of the database for the search engine index to identify the particular item and the predicted likelihood that was previously generated by the predictive model based at least on the one or more features associated with the subsequent search query;
selecting, using the predicted likelihood, the particular item of content; and
providing data identifying the particular item of content in response to the subsequent search query.
2. The method of claim 1 , wherein the one or more features relating to the current condition comprises a current time or a current event.
3. The method of claim 1 , wherein the one or more features relating to the device associated with the user comprises a location of the device, a configuration of the device, an operating system of the device, a screen resolution of the device, or a bandwidth of a connection to the device.
4. The method of claim 1 , wherein the one or more features relating to the user comprises a language used by the user, demographics of the user, or an account of the user.
5. The method of claim 1 , comprising:
training the predictive model using training data that associates a selection of an item of content and one or more corresponding features.
6. The method of claim 1 , wherein the predictive likelihood is further generated by applying weights to each of the one or more features.
7. The method of claim 1 , wherein the particular item of content is a video, document, advertisement, image, or an audio file.
8. The method of claim 1 , comprising:
determining the search query submitted by the user based on characteristics of a selection of an additional item of content, wherein the characteristics are related to an identity of the additional item of content or a topic of the additional item of content.
9. The method of claim 1 , wherein the search index is an inverted index or a posting list.
10. A system comprising:
one or more computers and one or more storage devices storing instructions that are operable, when executed by the one or more computers, to cause the one or more computers to perform operations comprising:
before receiving, at a search engine system, a subsequent search query that is associated with at least one or more features:
generating, using a predictive model and by the search engine system, data indicating a predicted likelihood of a user selecting a particular item of content after submitting a search query that is associated with the one or more features, wherein the particular item of content is a result for the search query, wherein the predicted likelihood is generated by the predictive model based at least on the one or more features, and wherein the one or more features include at least one feature relating to the user, the search query submitted by the user, the particular item of content, a device associated with the user, or a current condition; and
storing, by the search engine system and in a record entry of a database for a search engine index, data that associates the particular item of content and the one or more features with the predicted likelihood generated by the predictive model of the user selecting the particular item of content after submitting the search query that is associated with the one or more features;
receiving, at the search engine system, the subsequent search query that is associated with at least the one or more features;
in response to the subsequent search query:
accessing, by the search engine system, the record entry of the database for the search engine index to identify the particular item and the predicted likelihood that was previously generated by the predictive model based at least on the one or more features associated with the subsequent search query;
selecting, using the predicted likelihood, the particular item of content; and
providing data identifying the particular item of content in response to the subsequent search query.
11. The system of claim 10 , wherein the one or more features relating to the current condition comprises a current time or a current event.
12. The system of claim 10 , wherein the one or more features relating to the device associated with the user comprises a location of the device, a configuration of the device, an operating system of the device, a screen resolution of the device, or a bandwidth of a connection to the device.
13. The system of claim 10 , wherein the one or more features relating to the user comprises a language used by the user, demographics of the user, or an account of the user.
14. The system of claim 10 , wherein the operations further comprise:
training the predictive model using training data that associates a selection of an item of content and one or more corresponding features.
15. The system of claim 10 , wherein the predictive likelihood is further generated by applying weights to each of the one or more features.
16. The system of claim 10 , wherein the particular item of content is a video, document, advertisement, image, or an audio file.
17. The system of claim 10 , wherein the operations further comprise:
determining the search query submitted by the user based on characteristics of a selection of an additional item of content, wherein the characteristics are related to an identity of the additional item of content or a topic of the additional item of content.
18. The system of claim 10 , wherein the search index is an inverted index or a posting list.
19. A non-transitory computer-readable medium storing software comprising instructions executable by one or more computers which, upon such execution, cause the one or more computers to perform operations comprising:
before receiving, at a search engine system, a subsequent search query that is associated with at least one or more features:
generating, using a predictive model and by the search engine system, data indicating a predicted likelihood of a user selecting a particular item of content after submitting a search query that is associated with the one or more features, wherein the particular item of content is a result for the search query, wherein the predicted likelihood is generated by the predictive model based at least on the one or more features, and wherein the one or more features include at least one feature relating to the user, the search query submitted by the user, the particular item of content, a device associated with the user, or a current condition; and
storing, by the search engine system and in a record entry of a database for a search engine index, data that associates the particular item of content and the one or more features with the predicted likelihood generated by the predictive model of the user selecting the particular item of content after submitting the search query that is associated with the one or more features;
receiving, at the search engine system, the subsequent search query that is associated with at least the one or more features;
in response to the subsequent search query:
accessing, by the search engine system, the record entry of the database for the search engine index to identify the particular item and the predicted likelihood that was previously generated by the predictive model based at least on the one or more features associated with the subsequent search query;
selecting, using the predicted likelihood, the particular item of content; and
providing data identifying the particular item of content in response to the subsequent search query.
20. The medium of claim 19 , wherein the one or more features relating to the current condition comprises a current time or a current event.
21. The medium of claim 19 , wherein the one or more features relating to the device associated with the user comprises a location of the device, a configuration of the device, an operating system of the device, a screen resolution of the device, or a bandwidth of a connection to the device.
22. The medium of claim 19 , wherein the one or more features relating to the user comprises a language used by the user, demographics of the user, or an account of the user.
23. The medium of claim 19 , wherein the operations further comprise:
training the predictive model using training data that associates a selection of an item of content and one or more corresponding features.
24. The medium of claim 19 , wherein the predictive likelihood is further generated by applying weights to each of the one or more features.
25. The medium of claim 19 , wherein the operations further comprise:
determining the search query submitted by the user based on characteristics of a selection of an additional item of content, wherein the characteristics are related to an identity of the additional item of content or a topic of the additional item of content.