Haishan Liu

We describe a hybrid recommendation system at LinkedIn that seeks to optimally extract relevant information pertaining to items to be recommended. By extending the notion of an item profile, we propose the concept of a "virtual profile" that augments the content of the item with rich set of features inherited from members who have already shown explicit interest in it. Unlike item-based collaborative filtering, we focus on discovering the characteristic descriptors that underlie the item-user… Show more

We describe a hybrid recommendation system at LinkedIn that seeks to optimally extract relevant information pertaining to items to be recommended. By extending the notion of an item profile, we propose the concept of a "virtual profile" that augments the content of the item with rich set of features inherited from members who have already shown explicit interest in it. Unlike item-based collaborative filtering, we focus on discovering the characteristic descriptors that underlie the item-user association. Such information is used as supplemental features in a content-based filtering system. The main objective of virtual profiles is to provide a means to tap into rich-content information from one type of entity and propagate features extracted from which to other affiliated entities that may suffer from relative data scarcity. We empirically evaluate the proposed method on a real-world community recommendation problem at LinkedIn. The result shows that the virtual profiles outperform a collaborative filtering based approach (user who likes this also likes that). In particular, the improvement is more significant for new users with only limited connections, demonstrating the capability of the method to address the cold-start problem in pure collaborative filtering systems. Show less

In this paper, we present a data mining approach to address challenges in the matching of heterogeneous datasets. In particular, we propose solutions to two problems that arise in integrating information from different results of scientific research. The first problem, attribute matching, involves discovery of correspondences among distinct numeric features (attributes) that are used to characterize datasets that have been collected and analyzed in different research labs. The second problem… Show more

In this paper, we present a data mining approach to address challenges in the matching of heterogeneous datasets. In particular, we propose solutions to two problems that arise in integrating information from different results of scientific research. The first problem, attribute matching, involves discovery of correspondences among distinct numeric features (attributes) that are used to characterize datasets that have been collected and analyzed in different research labs. The second problem, cluster matching, involves discovery of matchings between patterns (clusters) across datasets. We treat both of these problems together as a multi-objective optimization problem. A multi-objective simulated annealing algorithm is described to find the optimal solution and compared with the genetic algorithm. The utility of this approach is demonstrated in a series of experiments using synthetic and realistic datasets that are designed to simulate heterogeneous data from different sources. Show less

In this paper, we address an interesting data mining problem of finding semantically associated item sets, i.e., items connected via indirect links. We propose a novel method for discovering semantically associated item sets based on a hyper graph representation of the database. We describe two similarity measures to compute the strength of associations between items. Specifically, we introduce the average commute time similarity, $\mathbf{s_{CT}}$, based on the random walk model on hyper… Show more

In this paper, we address an interesting data mining problem of finding semantically associated item sets, i.e., items connected via indirect links. We propose a novel method for discovering semantically associated item sets based on a hyper graph representation of the database. We describe two similarity measures to compute the strength of associations between items. Specifically, we introduce the average commute time similarity, $\mathbf{s_{CT}}$, based on the random walk model on hyper graph, and the inner-product similarity, $\mathbf{s_{L+}}$, based on the Moore-Penrose pseudoinverse of the hyper graph Laplacian matrix. Given semantically associated 2-itemsets generated by these measures, we design a hyper graph expansion method with two search strategies, namely, the clique and connected component search, to generate $k$-item sets ($k>2$). We show the proposed method is indeed capable of capturing semantically associated item sets through experiments performed on three datasets ranging from low to high dimensionality. The semantically associated item sets discovered in our experiment is promising to provide valuable insights on interrelationship between medical concepts and other domain specific concepts. Show less

In this paper, we present a data mining approach to chal- lenges in the matching and integration of heterogeneous datasets. In particular, we propose solutions to two problems that arise in combining information from different results of scientific research. The first problem, attribute matching, involves discovery of correspondences among distinct numeric-typed summary features (“attributes”) that are used to charac- terize datasets that have been collected and analyzed in different research… Show more

In this paper, we present a data mining approach to chal- lenges in the matching and integration of heterogeneous datasets. In particular, we propose solutions to two problems that arise in combining information from different results of scientific research. The first problem, attribute matching, involves discovery of correspondences among distinct numeric-typed summary features (“attributes”) that are used to charac- terize datasets that have been collected and analyzed in different research labs. The second problem, cluster matching, involves discovery of matchings between patterns across datasets. We treat both of these problems together as a multi-objective optimization problem. A multi-objective simulated annealing algorithm is described to find the optimal solution. The utility of this approach is demonstrated in a series of experiments using synthetic and realistic datasets that are designed to simulate heterogeneous data from different sources. Show less

Incorporating domain knowledge is one of the most challenging problems in data mining. The Semantic Web technologies are promising to offer solutions to formally capture domain knowledge and realize the efficient usage of the knowledge. We call the the data mining technology powered by the Semantic Web, capable of systematically incorporating domain knowledge, the semantic data mining. In this paper, we first survey a body of representative academic work that explores ontology- based… Show more

Incorporating domain knowledge is one of the most challenging problems in data mining. The Semantic Web technologies are promising to offer solutions to formally capture domain knowledge and realize the efficient usage of the knowledge. We call the the data mining technology powered by the Semantic Web, capable of systematically incorporating domain knowledge, the semantic data mining. In this paper, we first survey a body of representative academic work that explores ontology- based optimization for various data mining applications. Then we identify the semantic annotation as a crucial step towards semantic data mining since it brings meaning to data. Finally we propose a learning- based semantic search algorithm for annotating (semi-) structured data. Show less