Bignet Www2018

Apr. 30, 2018, most of slides (including both Keynotes and paper presentations) are online. See and download them at the Program section! Here are Dr. Jon Kleinberg's slides and Dr. Jure Leskovec's slides!

Mar. 28, 2018, BigNet@WWW'18 Program is released!

Mar. 03, 2018, Dr. Jon Kleinberg from Cornell joins BigNet@WWW'18 as a keynote speaker!

Feb. 26, 2018, Dr. Kuansan Wang from Microsoft Research joins BigNet@WWW'18 as a keynote speaker!

Feb. 25, 2018, Dr. Jure Leskovec from Stanford joins BigNet@WWW'18 as a keynote speaker!

Feb. 23, 2018, Dr. Paul Groth from LSW joins BigNet@WWW'18 as a keynote speaker!

Feb. 23, 2018, Drs. Michael Cochez, Mayank Kejriwal, Achim Rettinger from The Learning Semantics for the Web Workshop (LSW) join our BigNet workshop!

Feb. 14, 2018, nine papers are accepted from the BigNet@WWW'18 workshop!

Our world is networked: people are closer to each other through online social network services or mobile communication networks, while information is capable to be exchanged faster by World Wide Web or email networks. The network is a treasure trove of user experiences and knowledge that presents great opportunities to understand the fundamental science of our world. On the other hand, the network, with huge amount of data, and multiple types of entities (e.g., users, documents, organizations, etc.), user behaviors, and relations between entities, has become so large and complex that traditional methodologies are inadequate.

Over the last two decades, the conventional paradigm of mining and learning with networks usually starts from the explicit exploration of their structural properties. But many of such properties, such as betweenness centrality, triangle count, and modularity, require extensive domain knowledge and expensive computation to handcrafted. In light of these issues, as well as the opportunities offered by the recent emergence of representation learning, learning latent representations for networks, a.k.a., network embedding, has recently been extensively studied in order to automatically discover and map a network's structural properties into a latent space.

This workshop aims to provide a forum for presenting the most recent advances in network representation learning to unearth rich knowledge. We expect novel research works that address various aspects and challenges of this task, including learning representation for big networks, heterogeneous network embedding, embedding dynamic networks, scalable and efficient algorithms for graph embeddings, novel platforms and applications supporting network embeddings, comparisons between structural- and embedding-based network learning techniques, and beyond.

The WWW 2018 edition of BigNet invites leading experts in the area from all over the world. Thus, it serves also as a networking event both for connecting researchers from diverse research areas such as algorithms, data mining, and machine learning, and for connecting researchers from different geographic regions.

Jon Kleinberg, Cornell University, USA

Title: Link Prediction with Combinatorial Structure: Block Models and Simplicial Complexes [Slides]

Abstract: We show how the link prediction problem can be used to explore new kinds of combinatorial structure in complex network data. We first investigate methods based on enumerating walks of different lengths through a network, starting from the observation that a number of the most widely used techniques for link prediction, including personalized PageRank and heat kernel methods, operate in this space of enumerated walks. We show that for a naturally defined objective on graphs with stochastic block model structure, personalized PageRank emerges as the optimal function in a well-defined sense; this connection provides a novel formal motivation for the success of personalized PageRank in such contexts.

We then consider an extension of the link prediction problem to interactions that may take place among more than two nodes at once. Such higher-order interactions are ubiquitous in network data -- for example, communication within a group rather than person-to-person, collaboration among a team rather than a pair of co-authors, or biological interaction between a set of molecules rather than just two. We show how the link prediction problem provides a natural perspective for analyzing these higher-order interactions across a wide range of datasets, evaluating different approaches by how well they predict future multi-way interactions. Among our results, we find that different domains vary considerably in their distribution of higher-order interactions, and that our higher-order link prediction problem exhibits some fundamental differences from traditional pairwise link prediction, with a greater role for local rather than long-range information in predicting the appearance of new interactions.

This talk is based on joint work with Rediet Abebe, Austin Benson, Ali Jadbabaie, Isabel Kloumann, Michael Schaub, and Johan Ugander.

Bio: Jon Kleinberg is the Tisch University Professor in the Departments of Computer Science and Information Science at Cornell University. His research focuses on issues at the interface of algorithms, networks, and information, with an emphasis on the social and information networks that underpin the Web and other on-line media. He is a member of the US National Academy of Sciences and the US National Academy of Engineering, and the recipient of research fellowships from the MacArthur, Packard, Simons, and Sloan Foundations, as well as awards including the Harvey Prize, the Nevanlinna Prize, the SIGKDD Innovation Award, and the ACM Prize in Computing.

Kuansan Wang, Microsoft Research Redmond, USA

Title: Opportunities and Challenges of Vector Space Model for Web Scale Knowledge Representation

Abstract: Models that represent natural language (NL) semantics in a vector or Hilbert space have been proposed at the dawn of the computer age since 1940’s. The central idea, often referred to as distributional similarity, is based on a hypothesis that NL tokens that appear in similar context are more likely to be similar in meaning. The models have seen bursts of popularities over the past half of centuries, and only to be proven effective in recent years when massive amount of data become available and artificial neural network styles of nonlinearity are introduced. Motivated by the success, many have extended the idea of distributional similarity to heterogeneous knowledge graphs where the nodes and edges can be represented in high dimensional vectors with desired algebraic properties. In this talk, we will survey a few most notable examples, and demonstrate these approaches all share the same theoretical foundation and can be unified into a generalized framework. We further show the framework can be computed efficiently for a web scale knowledge graph and demonstrate a few application examples in Open Academic Graph.

Bio: Kuansan Wang is Managing Director and Principal Researcher from Microsoft Research Outreach in Redmond, WA. He joined Microsoft Research in 1998, first as a researcher in the Speech Technology Group working on multimodal dialog system, then as an architect that designed and shipped various speech products, including the Voice Command on mobile that eventually becomes Cortana, and Microsoft Speech Server that is still powering Microsoft and partners' call centers. In 2007, he rejoined Microsoft Research to work on large scale natural language understanding and web search technologies, and is currently responsible for running the largest machine reading efforts that use AI agents to dynamically acquire knowledge from the web and make it available to the general public. Kuansan received his BS from National Taiwan University and MS and PhD from University of Maryland, College Park, respectively, all in Electrical Engineering. In addition to 120+ scholarly papers and 40+ patents he has published, his work has also been adopted into 10 international standards from W3C, Ecma and ISO.

Jure Leskovec, Stanford University, USA

Title: Large-scale Graph Representation Learning [Slides]

Abstract: Machine learning on graphs is an important and ubiquitous task with applications ranging from drug design to friendship recommendation in social networks. The primary challenge in this domain is finding a way to represent, or encode, graph structure so that it can be easily exploited by machine learning models. However, traditionally machine learning approaches relied on user-defined heuristics to extract features encoding structural information about a graph. In this talk I will discuss methods that automatically learn to encode graph structure into low-dimensional embeddings, using techniques based on deep learning and nonlinear dimensionality reduction. I will provide a conceptual review of key advancements in this area of representation learning on graphs, including random-walk based algorithms, and graph convolutional networks.

Paul Groth, Elsevier Labs, Netherlands

Title: Combining Explicit and Latent Web Semantics

Abstract: Many networks in particular knowledge graphs have specific well defined semantics associated with portions of them. Types within a knowledge graph, for example, can be explicitly organized in a hierarchy or relations may defined as holding between two types. For instance, the network of structured data on the web has 38 billion edges with 9 billion types. Explicit semantics, while often imperfectly applied, can provide important clues in understanding the semantics of a network. In this talk, I discuss recent results in using explicit semantics as expressed by ontologies to guide the learning of latent semantic representations. Additionally, I discuss the goal of learning representations that can support increasingly expressive reasoning that was the initial target of web semantic languages.

1. Ganesh Jawahar, Ayushi Dalmia and Manish Gupta. Towards Interpretation of Node Embeddings
2. Ryan Rossi, Rong Zhou and Nesreen Ahmed. Deep Inductive Network Representation Learning
3. Ivan Brugere and Tanya Berger-Wolf. Network Model Selection Using Task-Focused Minimum Description Length
4. Giang Nguyen, John Lee, Ryan Rossi, Nesreen Ahmed, Eunyee Koh and Sungchul Kim. Continuous-Time Dynamic Network Embeddings
5. Andriy Nikolov, Peter Haase, Daniel Herzig, Johannes Trame and Artem Kozlov. Combining RDF Graph Data and Embedding Models for an Augmented Knowledge Graph

Non-archived Presentations:

1. Anton Tsitsulin, Davide Mottin, Panagiotis Karras and Emmanuel Müller. VERSE: Versatile Graph Embeddings from Similarity Measures
2. Yupeng Gu, Yizhou Sun, Yanen Li and Yang Yang. Social Rank Regulated Large-scale Network Embedding
3. Chao Huang and Nitesh Chawla. Camel: Content award and meta-path augmented metric learning for author identification
4. Junghwan Kim, Haekyu Park, Ji-Eun Lee and U Kang. SIDE: Representation Learning in Signed Directed Networks
5. Chun-Hsiang Wang, Kang-Chun Fan, Chuan-Ju Wang and Ming-Feng Tsai. Representation Learning for Sentiment Lexicon Generation from User Reviews

The program of the BigNet 2018 workshop will focus the attention on presenting and discussing the state-of-the-art, open problems, challenges and latest models, techniques and algorithms in the field of network embedding. In particular, this workshop focuses on presenting and discussing the development and applications of representation learning algorithms on large-scale networks and graphs. In this context, topics of interest include but are not limited to:

• § Network representation learning theories and foundations;
• § Representation learning for big networks;
• § Representation learning for heterogeneous networks;
• § Representation learning for dynamic networks;
• § Representation learning across multiple networks;
• § Representation learning for graphs with text;
• § Knowledge base embedding;
• § Deep learning for networks;
• § Graph theories and network embeddings;
• § Visualization for network embeddings;
• § Efficient graph embedding algorithms;
• § Scalable graph embedding models and frameworks;
• § Novel network embedding applications;
• § Representation learning and traditional structural mining.
• § Graph kernels and similarity

Important Dates

• § Submission Deadline: Jan. 20, 2018 (New).
• § Notification Date: Feb. 14, 2018.
• § Camera-Ready Submission: Mar. 04, 2018
• § All deadlines are at 11:59PM Alofi Time
• § Workshop Date: April 24, 2018

Submission Guidelines

The BigNet 2018 workshop encourages submissions that present both original results and preliminary/existing work on network embedding. We explicitly welcome extended-abstract submissions to introduce preliminary and arXiv work on network representation learning and knowledge graph embedding, as well as recently-published research at top conferences. The extended abstracts can option to be not archived in the WWW Companion proceeding. Therefore, this workshop accepts both full papers (4 to 8 pages) for original results and extended abstracts (1 to 2 pages) for published or ongoing work. All submissions must conform to the WWW 2018 main conference submission format: the ACM SIG Proceedings template with a font size no smaller than 9pt.

Submission site at easychair: Closed

Workshop Co-Chairs

Jie Tang, Tsinghua University, China

Michalis Vazirgiannis, Ecole Polytechnique, France

Yuxiao Dong, Microsoft Research, Redmond

Fragkiskos D. Malliaros, CentraleSupelec, University of Paris-Saclay, France

Michael Cochez, Fraunhofer Institute for Applied Information Technology, Germany

Mayank Kejriwal, Information Sciences Institute, USC.

Achim Rettinger, Karlsruhe Institute of Technology, Germany

Julian Mcauley, University of California San Diego
Chenhao Tan, University of Colorado Boulder
Jian Tang, HEC Montreal & MILA
William Hamilton, Stanford University
Nikolaos Tziortziotis, LIX, Ecole Polytechnique
Giannis Nikolentzos, Athens University of Economics and Business
James Foulds, University of Maryland, Baltimore County
Ryan Rossi, Adobe Research
Baichuan Zhang, Facebook
Jiliang Tang, Michigan State University
Apostolos N. Papadopoulos, Aristotle University of Thessaloniki

Pedro Szekely, University of Southern California

Contact

Should you have any questions, please direct your questions to jietang at tsinghua.edu.cn.

BigNet'17 @ ACM WWW 2017, Perth, Australia

BigNet'16 @ ACM CIKM 2016, Indianapolis, USA

AXA data science chair, Ecole Polytechnique