Training Mixed-Objective Pointing Decoders for Block-Level Optimization in Search Recommendation
This work addresses the challenge of improving search engine recommendations for users by integrating multiple optimization parameters into a block-level approach, though it appears incremental as it builds on existing ranking methods.
The paper tackles the problem of optimizing search recommendation blocks by proposing an architecture that processes all candidate suggestions for a query to output a block, rather than scoring suggestions individually, and experiments with mixed-objective training to enforce multiple metrics like relevance and diversity.
Related or ideal follow-up suggestions to a web query in search engines are often optimized based on several different parameters -- relevance to the original query, diversity, click probability etc. One or many rankers may be trained to score each suggestion from a candidate pool based on these factors. These scorers are usually pairwise classification tasks where each training example consists of a user query and a single suggestion from the list of candidates. We propose an architecture that takes all candidate suggestions associated with a given query and outputs a suggestion block. We discuss the benefits of such an architecture over traditional approaches and experiment with further enforcing each individual metric through mixed-objective training.