Data Mining - Concepts and Techniques.

Matrix Factorization Techniques for Recommender Systems

Applications of link prediction in social networks: A review

A novel POI recommendation method based on trust relationship and spatial–temporal factors

Aspect-Based Sentiment Analysis (ABSA) is a promising approach to analyze consumer reviews at a high level of detail, where the opinion about each feature of the product or service is considered ABSA usually explores supervised inductive learning algorithms, which requires intense human effort for the labeling process In this paper, we investigate Cross-Domain Transfer Learning approaches, in which aspects already labeled in some domains can be used to support the aspect extraction of another domain where there are no labeled aspects Existing cross-domain transfer learning approaches learn classifiers from labeled aspects in the source domain and then apply these classifiers in the target domain, ie, two separate stages that may cause inconsistency due to different feature spaces To overcome this drawback, we present an innovative approach called CD-ALPHN (Cross-Domain Aspect Label Propagation through Heterogeneous Networks) First, we propose a heterogeneous network-based representation that combines different features (labeled aspects, unlabeled aspects, and linguistic features) from source and target domain as nodes in a single network Second, we propose a label propagation algorithm for aspect extraction from heterogeneous networks, where the linguistic features are used as a bridge for this propagation Our algorithm is based on a transductive learning process, where we explore both labeled and unlabeled aspects during the label propagation Experimental results show that the CD-ALPHN outperforms the state-of-the-art methods in scenarios where there is a high-level of inconsistency between the source and target domains — the most common scenario in real-world applications

Cross-domain aspect extraction for sentiment analysis: A transductive learning approach

Recommendation generation using personalized weight of meta-paths in heterogeneous information networks

Transductive classification using labeled and unlabeled objects in a heterogeneous information network for knowledge extraction is an interesting and challenging problem. Most of the real-world networks are heterogeneous in their natural setting and traditional methods of classification for homogeneous networks are not suitable for heterogeneous networks. In a heterogeneous network, various meta-paths connecting objects of the target type, on which classification is to be performed, make the classification task more challenging. The semantic of each meta-path would lead to the different accuracy of classification. Therefore, weight learning of meta-paths is required to leverage their semantics simultaneously by a weighted combination. In this work, we propose a novel meta-path based framework, HeteClass, for transductive classification of target type objects. HeteClass explores the network schema of the given network and can also incorporate the knowledge of the domain expert to generate a set of meta-paths. The regularization based weight learning method proposed in HeteClass is effective to compute the weights of symmetric as well as asymmetric meta-paths in the network, and the weights generated are consistent with the real-world understanding. Using the learned weights, a homogeneous information network is formed on target type objects by the weighted combination, and transductive classification is performed. The proposed framework HeteClass is flexible to utilize any suitable classification algorithm for transductive classification and can be applied on heterogeneous information networks with arbitrary network schema. Experimental results show the effectiveness of the HeteClass for classification of unlabeled objects in heterogeneous information networks using real-world data sets.

HeteClass: A Meta-path based framework for transductive classification of objects in heterogeneous information networks

The real world contains many complex networks, including research networks, social networks, biological networks, and transport networks. Real-world complex networks are unconstrained and can be characterized as undirected and unweighted. Understanding and controlling the process of information propagation in such networks is significant for decision-making activities and has many uses, such as disease control, market advertising, rumor control, and innovation propagation. Identifying the influencers in complex networks is an important activity, as influencers play a key role in spreading information to aid the decision-making process. In this study, we consider the problem of identifying a set of top-N influential nodes for spreading the information in undirected and unweighted networks using the network structure in the absence of domain-specific knowledge. In this study, we propose a novel method that computes the ranking scores of the nodes in the network and considers the influence of other nodes simultaneously when forming the set of top-N influential nodes. The proposed method is different from other methods of identification of influential nodes in the network, in that it takes into consideration the position of the nodes in the network while computing the ranking score, thereby preventing the clustering of important nodes, which hampers the information flow. Experiments are performed using several real-world complex networks to demonstrate the effectiveness of the proposed method.

Spreading the information in complex networks: Identifying a set of top-N influential nodes using network structure

Measuring relatedness between objects (nodes) in a heterogeneous network is a challenging and an interesting problem. Many people transform a heterogeneous network into a homogeneous network before applying a similarity measure. However, such transformation results in information loss as path semantics are lost. In this paper, we study the problem of measuring relatedness between objects in a heterogeneous network using only link information and propose a meta-path based novel measure for relevance measurement in a general heterogeneous network with a specified network schema. The proposed measure is semi-metric and incorporates the path semantics by following the specified meta-path. For relevance measurement, using the specified meta-path, the given heterogeneous network is converted into a bipartite network consisting only of source and target type objects between which relatedness is to be measured. In order to validate the effectiveness of the proposed measure, we compared its performance with existing relevance measures which are semi-metric and applicable to heterogeneous networks. To show the viability and the effectiveness of the proposed measure, experiments were performed on real world bibliographic dataset DBLP. Experimental results show that the proposed measure effectively measures the relatedness between objects in a heterogeneous network and it outperforms earlier measures in clustering and query task.

A New Relevance Measure for Heterogeneous Networks

Mining a heterogeneous network requires measuring the relatedness between objects represented as nodes in the network. Relevance measurement on objects in a heterogeneous network is a challenging problem. Many researchers transform a heterogeneous network into the corresponding homogeneous network and then apply conventional similarity measures. However, this approach involves information loss as various path semantics are lost in the transformation process. In this paper, we study the problem of relevance measurement on objects in a heterogeneous network and propose a meta-path based semi-metric measure for relevance measurement on objects in a general heterogeneous network with a specified network schema. The proposed measure incorporates path semantics by following the specified meta-path. For measuring relatedness between objects using the proposed measure, the heterogeneous network is converted into a bipartite network consisting of source and target type objects following the specified meta-path. To validate the effectiveness of the proposed measure, we compare its performance with the existing meta-path based semi-metric measures applicable to heterogeneous networks. Experiments are performed on real-world datasets to show the effectiveness of the proposed measure.

Mukul Gupta

Papers

Recommendation generation using personalized weight of meta-paths in heterogeneous information networks

HeteClass: A Meta-path based framework for transductive classification of objects in heterogeneous information networks

Spreading the information in complex networks: Identifying a set of top-N influential nodes using network structure

A New Relevance Measure for Heterogeneous Networks

DPRel: A Meta-Path Based Relevance Measure for Mining Heterogeneous Networks