Improving Retrieval in Theme-specific Applications using a Corpus Topical Taxonomy
This work addresses retrieval challenges in specialized domains like industries or niche areas, but it is incremental as it builds on existing PLM-based retrievers with a plug-and-play enhancement.
The paper tackles the problem of limited effectiveness of large-scale pre-trained language models in theme-specific document retrieval due to unique terminologies and incomplete query contexts, and proposes the ToTER framework that uses a corpus topical taxonomy to improve retrieval by identifying central topics and exploiting topical relatedness, demonstrating effectiveness through experiments on two real-world datasets.
Document retrieval has greatly benefited from the advancements of large-scale pre-trained language models (PLMs). However, their effectiveness is often limited in theme-specific applications for specialized areas or industries, due to unique terminologies, incomplete contexts of user queries, and specialized search intents. To capture the theme-specific information and improve retrieval, we propose to use a corpus topical taxonomy, which outlines the latent topic structure of the corpus while reflecting user-interested aspects. We introduce ToTER (Topical Taxonomy Enhanced Retrieval) framework, which identifies the central topics of queries and documents with the guidance of the taxonomy, and exploits their topical relatedness to supplement missing contexts. As a plug-and-play framework, ToTER can be flexibly employed to enhance various PLM-based retrievers. Through extensive quantitative, ablative, and exploratory experiments on two real-world datasets, we ascertain the benefits of using topical taxonomy for retrieval in theme-specific applications and demonstrate the effectiveness of ToTER.