SimRank

About: SimRank is a research topic. Over the lifetime, 250 publications have been published within this topic receiving 21163 citations.

Efficient Similarity Search for Travel Behavior

Abstract: To provide travel recommendations and planning in the intelligent transportation system (ITS), we must have the ability to find similar travel patterns among users based on their real mobility traces. To measure the similarity of user’s travel behavior, various methods have been proposed, but they usually only rely on a single attributes-related metric. In comparison, studies of the semantic relationships between travel attributes remain scarce, making it difficult to construct a complete mobility pattern that reveals the relevance between users or groups. In this paper, we introduced the heterogeneous information network to build a weighted travel network with spatial–temporal GPS trajectories. The heterogeneous network allows clustering the similar users based on the connections between different attributes instead of attribute values. On this basis, we defined the meta-paths for travel and used each meta-path to formulate a similarity measure over users by improving existing PathSim (Meta-path-based similarity measures) and SimRank. Next, we aggregated different similarities, where each meta-path was automatically weighted by the learning algorithm to make predictions. The experimental results showed that the recall of the similarity measurement algorithm using multiple meta-paths has improved, which yielded better results than the performance of the algorithm using a single meta-path. The performance of the improved PathSim model under different scales of data was 15% higher than the performance of the improved SimRank model in terms of precision and 21% higher in terms of recall. Due to the area under curve values, our experiments also show that a meta-path combination is more effective than the state-of-the-art approaches and can be efficiently computed.

Tutorial: Lessons Learned from Building Real-life Recommender Systems

Abstract: In 2006, Netflix announced a \$1M prize competition to advance recommendation algorithms. The recommendation problem was simplified as the accuracy in predicting a user rating measured by the Root Mean Squared Error. While that formulation helped get the attention of the research community, it put the focus on the wrong approach and metric while leaving many important factors out. In this tutorial we will describe the advances in Recommender Systems in the last 10 years from an industry perspective based on the instructors' personal experience at companies like Quora, LinkedIn, Netflix, or Yahoo! We will do so in the form of different lessons learned through the years.Some of those lessons will describe the different components of modern recommender systems such as: personalized ranking, similarity, explanations, context-awareness, or multi-armed bandits. Others will also review the usage of novel algorithmic approaches such as Factorization Machines, Restricted Boltzmann Machines, SimRank, Deep Neural Networks, or Listwise Learning-to-rank. Others will dive into details of the importance of gathering the right data or using the correct optimization metric.But, most importantly, we will give many examples of prototypical industrial-scale recommender systems with special focus on those unsolved challenges that should define the future of the recommender systems area.

Selecting related terms in query-logs using two-stage SimRank

Abstract: It is commonly believed that query logs from Web search are a gold mine for search business, because they reflect users' preference over Web pages presented by search engines, so a lot of studies based on query logs have been carried out in the last few years. In this study, we assume that two queries are relevant to each other when they have same clicked page in their result lists, and we also consider the queries' topics of user's need. Thus, we propose a Two-Stage SimRank (called TSS in this paper) algorithm based on SimRank and some clustering algorithms to compute the similarity among queries, and then use it to discover relevant terms for query expansion, considering the information of topics and the global relationships of queries concurrently, with a query log collected by a practical search engine. Experimental results on two TREC test collections show that our approach can discover qualified terms effectively and improve retrieval performance.

SimRank: A link analysis based blogger recommendation algorithm using text similarity

Abstract: Blogs have influenced our life profoundly and people preferred to subscribe to influential bloggers they are interested in. Hence, the identification of influential bloggers automatically has become an important task. Previous researches focus on the blog sites in which there exist abundant hyperlinks, but their methods can not scale to ones that have few hyperlinks. In this paper, we propose a novel algorithm called SimRank to recommend influential bloggers. Our algorithm is based on the observation that the reproduction of blog posts and similar contents is common in blogosphere, which form implicit links between bloggers. By measuring the text similarity of the blog posts, we create link graph between bloggers, and adopt the PageRank algorithm to rank the importance of bloggers. Experimental results indicate that our proposed algorithm is effective in recommending bloggers.

An Adaptive Method for the Efficient Similarity Calculation

Abstract: SimRank is a well-known algorithm for similarity calculation based on object-to-object relationship. However, it suffers from high computation cost. In this paper, we find that the convergence behavior of different object pairs is different when we use SimRank to compute the similarity of objects. Many similarity scores converge fast, while others need more time before convergence. Based on this observation, we propose an adaptive method called Adaptive-SimRank to speed up similarity calculation. Using this method, we don't need to recalculate those converged pairs' similarity. The experiments conducted on web datasets and synthetic dataset show that our new method can reduce the running time by nearly 35%.