Search results

Learning to rank algorithms inherently faces many challenges. The most important challenges could be listed as high-dimensionality of the training data, the dynamic nature of Web information resources and lack of click-through data. High dimensionality of the training data affects effectiveness and efficiency of learning algorithms. Besides, most of learning to rank benchmark datasets do not include click-through data as a very rich source of information about the search behavior of users while dealing with the ranked lists of search results. To deal with these limitations, this paper aims to introduce a novel learning to rank algorithm by using a set of complex click-through features in a reinforcement learning (RL) model. These features are calculated from the existing click-through information in the data set or even from data sets without any explicit click-through information.

The proposed ranking algorithm (QRC-Rank) applies RL techniques on a set of calculated click-through features. QRC-Rank is as a two-steps process. In the first step, Transformation phase, a compact benchmark data set is created which contains a set of click-through features. These feature are calculated from the original click-through information available in the data set and constitute a compact representation of click-through information. To find most effective click-through feature, a number of scenarios are investigated. The second phase is Model-Generation, in which a RL model is built to rank the documents. This model is created by applying temporal difference learning methods such as Q-Learning and SARSA.

The proposed learning to rank method, QRC-rank, is evaluated on WCL2R and LETOR4.0 data sets. Experimental results demonstrate that QRC-Rank outperforms the state-of-the-art learning to rank methods such as SVMRank, RankBoost, ListNet and AdaRank based on the precision and normalized discount cumulative gain evaluation criteria. The use of the click-through features calculated from the training data set is a major contributor to the performance of the system.

In this paper, we have demonstrated the viability of the proposed features that provide a compact representation for the click through data in a learning to rank application. These compact click-through features are calculated from the original features of the learning to rank benchmark data set. In addition, a Markov Decision Process model is proposed for the learning to rank problem using RL, including the sets of states, actions, rewarding strategy and the transition function.

Incorporating users’ behavior patterns could help in the ranking process. Different click models (CMs) are introduced to model the sophisticated search-time behavior of users among which commonly used the triple of attractiveness, examination and satisfaction. Inspired by this fact and considering the psychological definitions of these concepts, this paper aims to propose a novel learning to rank by redefining these concepts. The attractiveness and examination factors could be calculated using a limited subset of information retrieval (IR) features by the random forest algorithm, and then they are combined with each other to predicate the satisfaction factor which is considered as the relevance level.

The attractiveness and examination factors of a given document are usually considered as its perceived relevance and the fast scan of its snippet, respectively. Here, attractiveness and examination factors are regarded as the click-count and the investigation rate, respectively. Also, the satisfaction of a document is supposed to be the same as its relevance level for a given query. This idea is supported by the strong correlation between attractiveness-satisfaction and the examination-satisfaction. Applying random forest algorithm, the attractiveness and examination factors are calculated using a very limited set of the primitive features of query-document pairs. Then, by using the ordered weighted averaging operator, these factors are aggregated to estimate the satisfaction.

Experimental results on MSLR-WEB10K and WCL2R data sets show the superiority of this algorithm over the state-of-the-art ranking algorithms in terms of P@n and NDCG criteria. The enhancement is more noticeable in top-ranked items which are reviewed more by the users.

This paper proposes a novel learning to rank based on the redefinition of major building blocks of the CMs which are the attractiveness, examination and satisfactory. It proposes a method to use a very limited number of selected IR features to estimate the attractiveness and examination factors and then combines these factors to predicate the satisfactory which is regarded as the relevance level of a document with respect to a given query.