ABLE: Meta-Path Prediction in Heterogeneous Information Networks

2022 ◽  
Vol 16 (4) ◽  
pp. 1-21
Author(s):  
Chenji Huang ◽  
Yixiang Fang ◽  
Xuemin Lin ◽  
Xin Cao ◽  
Wenjie Zhang
Keyword(s):  
State Of The Art ◽  
Information Networks ◽  
Information Network ◽  
Embedding Method ◽  
Heterogeneous Information ◽  
Heterogeneous Information Networks ◽  
Head Node ◽  
Meta Path ◽  
Path Prediction ◽  
Ap Scores

Given a heterogeneous information network (HIN) H, a head node h , a meta-path P, and a tail node t , the meta-path prediction aims at predicting whether h can be linked to t by an instance of P. Most existing solutions either require predefined meta-paths, which limits their scalability to schema-rich HINs and long meta-paths, or do not aim at predicting the existence of an instance of P. To address these issues, in this article, we propose a novel prediction model, called ABLE, by exploiting the A ttention mechanism and B i L STM for E mbedding. Particularly, we present a concatenation node embedding method by considering the node types and a dynamic meta-path embedding method that carefully considers the importance and positions of edge types in the meta-paths by the Attention mechanism and BiLSTM model, respectively. A triplet embedding is then derived to complete the prediction. We conduct extensive experiments on four real datasets. The empirical results show that ABLE outperforms the state-of-the-art methods by up to 20% and 22% of improvement of AUC and AP scores, respectively.

Meta-Path Based Inductive Classification in Heterogeneous Information Networks

2016 ◽  
Vol 13 (10) ◽  
pp. 6747-6753
Author(s):  
Pingjian Ding ◽  
Xiangtao Chen ◽  
Zipin Guan
Keyword(s):  
Heterogeneous Networks ◽  
State Of The Art ◽  
Test Sample ◽  
Information Networks ◽  
Classification Problems ◽  
Heterogeneous Information ◽  
Heterogeneous Information Networks ◽  
Meta Path ◽  
Transductive Inference ◽  
The Given

The goal of inductive classification approaches is to infer the correct mapping from test set to labels, while the goal of transductive inference is to predict the correct labels for the given unlabeled data. Hence, the increased unlabeled samples can’t be classified by transductive classification. In this paper, we focus on studying the inductive classification problems in heterogeneous networks, which involve multiple types of objects interconnected by multiple types of links. Moreover, the objects and the links are gradually increasing over time. To accommodate characteristics of heterogeneous networks, a meta-path-based heterogeneous inductive classification (Hic) was proposed. First, the different sub-networks were constructed according to the selected meta-path. Second, the characteristic paths of each sub-network were extracted via the specified minimum support, and were assigned appropriate weights. Then, Hic model based on characteristic path was built. Finally, the Hic scores of each classification label for each test sample was calculated via links between test samples and sub-networks. Experiments on the DBLP showed that the proposed method significantly improves the accuracy and stability over the existing state-of-the-art methods for classification in dynamic heterogeneous network.

Distributed Algorithms for Finding Meta-Paths of a Large Heterogeneous Information Network on Cloud

2020 ◽  
pp. 223-249
Author(s):  
Phuc Do
Keyword(s):  
Decision Making ◽  
Distributed Algorithms ◽  
Information Networks ◽  
Information Network ◽  
Breadth First Search ◽  
Cloud Computing Environment ◽  
Heterogeneous Information ◽  
Heterogeneous Information Networks ◽  
Path Discovery ◽  
Meta Path

Meta-path is an important concept of heterogeneous information networks (HINs). Meta-paths were used in many tasks such as information retrieval, decision making, and product recommendation. Normally meta-paths were proposed by human experts. Recently, works on meta-path discovery have proposed in-memory solutions that fit in one computer. With large HINs, the whole HIN cannot be loaded in the memory. In this chapter, the authors proposed distributed algorithms to discover meta-paths of large HINs on cloud. They develop the distributed algorithms to discover the significant meta-path, maximal significant meta-path, and top-k meta-paths between two vertices of HIN. Calculation of the support of meta-paths or performing breadth first search can be computational costly in very large HINs. Conveniently, the distributed algorithms utilize the GraphFrames library of Apache Spark on cloud computing environment to efficiently query large HINs. The authors conduct the experiments on large DBLP dataset to prove the performance of our algorithms on cloud.

scholarly journals A Survey of Data Mining Techniques on Information Networks

2018 ◽  
Vol 7 (2.6) ◽  
pp. 293
Author(s):  
Sadhana Kodali ◽  
Madhavi Dabbiru ◽  
B Thirumala Rao
Keyword(s):  
Data Mining ◽  
Social Media ◽  
Information Networks ◽  
Information Network ◽  
Heterogeneous Information ◽  
Data Mining Techniques ◽  
Heterogeneous Information Networks ◽  
Web Objects ◽  
The Social ◽  
Social Media Network

An Information Network is the network formed by the interconnectivity of the objects formed due to the interaction between them. In our day-to-day life we can find these information networks like the social media network, the network formed by the interaction of web objects etc. This paper presents a survey of various Data Mining techniques that can be applicable to information networks. The Data Mining techniques of both homogeneous and heterogeneous information networks are discussed in detail and a comparative study on each problem category is showcased.

A comparative study on heterogeneous information network embeddings

2020 ◽  
Vol 39 (3) ◽  
pp. 3463-3473
Author(s):  
Fujiao Ji ◽  
Zhongying Zhao ◽  
Hui Zhou ◽  
Heng Chi ◽  
Chao Li
Keyword(s):  
Comparative Study ◽  
Communication Networks ◽  
Problem Definition ◽  
Information Networks ◽  
Future Research ◽  
Information Network ◽  
Network Embedding ◽  
Heterogeneous Information Network ◽  
Heterogeneous Information ◽  
Heterogeneous Information Networks

Heterogeneous information networks are widely used to represent real world applications in forms of social networks, word co-occurrence networks, and communication networks, etc. However, It is difficult for traditional machine learning methods to analyze these networks effectively. Heterogeneous information network embedding aims to convert the network into low dimensional vectors, which facilitates the following tasks. Thus it is receiving tremendous attention from the research community due to its effectiveness and efficiency. Although numerous methods have been present and applied successfully, there are few works to make a comparative study on heterogeneous information network embedding, which is very important for developers and researchers to select an appropriate method. To address the above problem, we make a comparative study on the heterogeneous information network embeddings. Specifically, we first give the problem definition of heterogeneous information network embedding. Then the heterogeneous information networks are classified into four categories from the perspective of network type. The state-of-the-art methods for each category are also compared and reviewed. Finally, we make a conclusion and suggest some potential future research directions.

Sign in / Sign up

Export Citation Format

Share Document