Abstract
Information retrieval systems can face challenges interpreting a user's intent during follow-up queries, as relying on prior query history alone may be insufficient to resolve ambiguity. Systems and methods can address this by analyzing the semantic structure of a document a user is currently viewing, such as its document object model. When a user selects text to initiate a follow-up request, the system can extract contextual information, including surrounding text and hierarchical elements like section headings. This structural context, along with the user's initial query and the selected text, may be provided to a generative model to synthesize a new, detailed query. This process can generate a query that more accurately reflects the user's specific informational need within their current reading context, which may improve the relevance of retrieved information for the follow-up request.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Miglino, Jerome; Zhang, Zhuowei; Herrick, Corinn; Bolelli, Levent; Mehta, Pallav; Leszczuk, Will; and Shi, MJ, "Contextual Query Generation Using Document Semantic Structure", Technical Disclosure Commons, (March 11, 2026)
https://www.tdcommons.org/dpubs_series/9498