Beyond [CLS] through Ranking by Generation
This work addresses the problem of improving ranking accuracy in information retrieval for users needing precise answer selection, though it is incremental as it revisits and updates an existing framework.
The authors revisited generative models for information retrieval, showing that their generative approaches are as effective as state-of-the-art discriminative models for answer selection, and demonstrated the effectiveness of unlikelihood losses in this context.
Generative models for Information Retrieval, where ranking of documents is viewed as the task of generating a query from a document's language model, were very successful in various IR tasks in the past. However, with the advent of modern deep neural networks, attention has shifted to discriminative ranking functions that model the semantic similarity of documents and queries instead. Recently, deep generative models such as GPT2 and BART have been shown to be excellent text generators, but their effectiveness as rankers have not been demonstrated yet. In this work, we revisit the generative framework for information retrieval and show that our generative approaches are as effective as state-of-the-art semantic similarity-based discriminative models for the answer selection task. Additionally, we demonstrate the effectiveness of unlikelihood losses for IR.