Optimal Top-k Document Retrieval

TL;DR

This paper introduces a linear-space data structure for optimal top-k document retrieval that supports various relevance measures, dynamic updates, and range-restricted queries, with efficient query and update times.

Contribution

It presents a novel RAM-optimal suffix tree search method and extends top-k retrieval to dynamic and range-restricted scenarios with improved space and time complexities.

Findings

Achieves optimal query time in the RAM model for static retrieval.

Supports dynamic document insertion and deletion with efficient update times.

Provides solutions for range-restricted top-k queries with linear space.

Abstract

Let $\mathcal{D}$ be a collection of $D$ documents, which are strings over an alphabet of size $\sigma$, of total length $n$. We describe a data structure that uses linear space and and reports $k$ most relevant documents that contain a query pattern $P$, which is a string of length $p$, in time $O(p/\log_\sigma n+k)$, which is optimal in the RAM model in the general case where $\lg D = \Theta(\log n)$, and involves a novel RAM-optimal suffix tree search. Our construction supports an ample set of important relevance measures... [clip] When $\lg D = o(\log n)$, we show how to reduce the space of the data structure from $O(n\log n)$ to $O(n(\log\sigma+\log D+\log\log n))$ bits... [clip] We also consider the dynamic scenario, where documents can be inserted and deleted from the collection. We obtain linear space and query time $O(p(\log\log n)^2/\log_\sigma n+\log n + k\log\log k)$, whereas insertions and deletions require $O(\log^{1+\epsilon} n)$ time per symbol, for any constant $\epsilon>0$. Finally, we consider an extended static scenario where an extra parameter $par(P,d)$ is defined, and the query must retrieve only documents $d$ such that $par(P,d)\in [\tau_1,\tau_2]$, where this range is specified at query time. We solve these queries using linear space and $O(p/\log_\sigma n + \log^{1+\epsilon} n + k\log^\epsilon n)$ time, for any constant $\epsilon>0$. Our technique is to translate these top-$k$ problems into multidimensional geometric search problems. As an additional bonus, we describe some improvements to those problems.