Recency Ranking by Diversification of Result Set

TL;DR

The paper tackles recency-sensitive search by treating freshness as a non-topical facet and applying result set diversification to blend fresh content into ordinary rankings. It builds a regression-based classifier to estimate the probability that a query requires recent information and optimizes an Extended ERR-IAA objective to maximize user satisfaction under temporal ambiguity. Key contributions include the recency-sensitive query classifier, a diversification framework with a formal ERR-IAA formulation, and large-scale offline/online evaluations showing improved engagement and speeded interactions in real-world search traffic. The approach offers a practical pathway to time-aware retrieval, potentially extendable to other verticals while highlighting considerations around freshness windows and potential over-diversification.

Abstract

In this paper, we propose a web search retrieval approach which automatically detects recency sensitive queries and increases the freshness of the ordinary document ranking by a degree proportional to the probability of the need in recent content. We propose to solve the recency ranking problem by using result diversification principles and deal with the query's non-topical ambiguity appearing when the need in recent content can be detected only with uncertainty. Our offline and online experiments with millions of queries from real search engine users demonstrate the significant increase in satisfaction of users presented with a search result generated by our approach.

Figures (4)

• Figure 1: Recency sensitive queries traffic coverage
• Figure 2: Cumulative share of query instances submitted since the day of the first query
• Figure 3: ERR-IAA for queries with different "true" probabilities of need in fresh content
• Figure 4: ERR-IAA for queries with different level of ambiguity, diversified and non-diversified rankings