Abstract: An online system selects an item to present to a user of the online system. The online system accesses user interaction data for the user. The online system transmits the user interaction data to a model serving system and receives, from the model serving system, item embeddings for the items with which the user interacted. The model serving system may use an LLM to generate the item embeddings based on the user interaction data. The online system generates a user embedding array based on the item embeddings. The online system applies a transformer network to the user embedding array to generate a user embedding describing the user. To select an item to present to the user, the online system compares the generated user embedding to item embeddings for a set of candidate items. The online system selects a candidate item based on the interaction scores.

Abstract: An online system accesses a two-tower model trained to identify candidate items for presentation to users, in which the model includes an item tower trained to compute item embeddings and a user tower trained to compute user embeddings. The user tower includes a long-term sub-tower trained to compute long-term embeddings for users and a short-term sub-tower trained to compute short-term embeddings for users. The model is trained based on item data associated with items, user data associated with users, and session data associated with user sessions. The system uses the item tower to compute an item embedding for each of multiple candidate items. The system also uses the long-term sub-tower to compute a long-term embedding for a user. The system then receives session data associated with a current session of the user and uses the short-term sub-tower to compute a short-term embedding for the user based on this session data.

Abstract: Systems, methods, and non-transitory computer-readable media can receive a user query comprising one or more search terms. One or more synonyms are identified for the user query based on a dynamic thesaurus generated using automated synonym extraction. An expanded query is generated based on the user query and the one or more synonyms. One or more search results are identified based on the expanded query.

Abstract: Systems, methods, and non-transitory computer-readable media can receive a request from a user associated with a particular location to receive location-based notifications for the particular location. A subscription associated with the user and the particular location is generated. One or more pages associated with the particular location are identified. A location-based notification comprising the one or more pages is generated for the user.

Abstract: Systems, methods, and non-transitory computer readable media can obtain a plurality of page engagement graphs, each of the plurality of page engagement graphs associated with a page engagement type of a plurality of page engagement types. Respective weights associated with the plurality of page engagement types can be determined. An aggregated page engagement graph can be generated based on the plurality of page engagement graphs and the respective weights. Pages in the aggregated page engagement graph can be ranked.

Abstract: Systems, methods, and non-transitory computer-readable media can compute a query embedding in a first multi-dimensional space based on a query embedding model. The query embedding is associated with a user query. A plurality of page embeddings are computed in a second multi-dimensional space based on a page embedding model. A query joint embedding and a plurality of page joint embeddings are computed in a third multi-dimensional space based on the query embedding, the plurality of page embeddings, and a joint embedding model. One or more page results are identified for the user query based on the query joint embedding and the plurality of page joint embeddings.