Query reranking as a service

Abstract

The ranked retrieval model has rapidly become the de facto way for search query processing in client-server databases, especially those on the web. Despite of the extensive efforts in the database community on designing better ranking functions/mechanisms, many such databases in practice still fail to address the diverse and sometimes contradicting preferences of users on tuple ranking, perhaps (at least partially) due to the lack of expertise and/or motivation for the database owner to design truly effective ranking functions. This paper takes a different route on addressing the issue by defining a novel query reranking problem, i.e., we aim to design a thirdparty service that uses nothing but the public search interface of a client-server database to enable the on-the-fly processing of queries with any user-specified ranking functions (with or without selection conditions), no matter if the ranking function is supported by the database or not. We analyze the worst-case complexity of the problem and introduce a number of ideas, e.g., on-the-fly indexing, domination detection and virtual tuple pruning, to reduce the average-case cost of the query reranking algorithm. We also present extensive experimental results on real-world datasets, in both offline and live online systems, that demonstrate the effectiveness of our proposed techniques.

On obtaining stable rankings

“…Given a dataset with multiple attributes, ordering the items and choosing a subset to support decision making is challenging. This has motivated a rich body of work on ranking [23,[45][46][47][48], topk [21,22], and skyline queries [8,10,49,50]. Broadly speaking, ranking and top-k are employed when a user's preference in the form of a scoring function is available, while skyline queries are used when only the scoring attributes are known, but the scoring function is left unspecified.…”

QR2: A Third-Party Query Reranking Service over Web Databases

“…In addition to the technical challenges addressed in [11], the following are the main challenges we faced in practice, together with their resolutions:…”

“…• General positioning assumption: a general assumption in [11] is that no two tuples have the same values on a given attribute. This assumption, however, may not hold in practice.…”

“…As a result, there is often a significant gap, in terms of both design and diversity, between the ranking function(s) supported by the web database and the true preferences of the database users. While ranked retrieval model [1], [2], top-k query processing [3], [4], and building indices for them [5], [6] on one side, and hidden web databases [7], [8], [9] on the other side are well studied in the literature, research on ranked retrieval model on web databases is limited to the two recent papers [10], [11]. Similar to [11], [12] also proposes a thirdparty service on web databases, however unlike [11], it focuses on faceted search for mobile applications.…”

“…In this paper, we 1 www.zillow.com 2 www.bluenile.com demonstrate QR2, a working solution based on [11], a thirdparty service that uses nothing but the public search interface of a web database to enable the on-the-fly processing of queries with any user-specified ranking functions. Proposed algorithms in [11] enable the "Get-Next" primitive that provide an incremental reranking of the results. Problem definition: Consider a web database D with a top-k interface and an arbitrary, unknown, system ranking function.…”