Modeling Queries as Bags of Documents
Last week, I had the honor of presenting “Modeling Queries as Bags of Documents” at Search Solutions 2024 with Aritra Mandal.
We introduced the bag-of-documents model as a way to align query and document representations — specifically addressing the gap between the broad variability of query intents and the inherent specificity of individual documents or products. We described how to compute bag-of-documents representations of frequent queries by aggregating document vectors from their clicks and then using those query vectors as training data to build a sentence transformer model for infrequent queries. We then showed how the bag-of-documents model is useful to recognize query similarity and compute query specificity, both of which are helpful for improving quality, experience, and analytics for search applications.
Here are the slides. Enjoy!
