Design Patterns for Fusion-Based Object Retrieval
This work provides a general framework for object retrieval, addressing a specific challenge in information retrieval for tasks like expert finding, but it is incremental as it organizes existing approaches rather than introducing new methods.
The paper tackles the problem of ranking objects without direct term-based representations by introducing two design patterns for fusion-based retrieval strategies, early and late fusion, and demonstrates their generality across expert finding, blog distillation, and vertical ranking tasks.
We address the task of ranking objects (such as people, blogs, or verticals) that, unlike documents, do not have direct term-based representations. To be able to match them against keyword queries, evidence needs to be amassed from documents that are associated with the given object. We present two design patterns, i.e., general reusable retrieval strategies, which are able to encompass most existing approaches from the past. One strategy combines evidence on the term level (early fusion), while the other does it on the document level (late fusion). We demonstrate the generality of these patterns by applying them to three different object retrieval tasks: expert finding, blog distillation, and vertical ranking.