Diversifying search results to satisfy as many users' intentions as possible is NP-hard. Some research employs a pruned exhaustive search, and some uses a greedy approach. However, the objective function of the result diversification problem adopts the cascade assumption which assumes users' information needs will drop once their subtopic search intents are satisfied. As a result, the intent distribution of diversified results deviates from the actual distribution of user intentions, and each subtopic tends to be chosen equally. This phenomenon is unreasonable, especially when the original distribution of user intent is unbalanced. In this paper, we present empirical evidence of the diversification equilibrium by showing that the standard deviations of subtopic distribution approach zero.
[1]
Yiqun Liu,et al.
A pruning algorithm for optimal diversified search
,
2014,
WWW '14 Companion.
[2]
M. L. Fisher,et al.
An analysis of approximations for maximizing submodular set functions—I
,
1978,
Math. Program..
[3]
Sreenivas Gollapudi,et al.
Diversifying search results
,
2009,
WSDM '09.
[4]
Craig MacDonald,et al.
On the role of novelty for search result diversification
,
2011,
Information Retrieval.
[5]
David R. Karger,et al.
Less is More Probabilistic Models for Retrieving Fewer Relevant Documents
,
2006
.
[6]
Fabrizio Silvestri,et al.
Efficient Diversification of Web Search Results
,
2011,
Proc. VLDB Endow..