This year, for the first time in its 26-year history, The Web Conference (formerly known as WWW) will be entirely virtual, streaming interactively over both the Zoom and Amazon Chime platforms.
A number of computer science conferences have adopted similar measures in the face of the COVID-19 pandemic, but The Web Conference is one of the first — and one of the largest.
“The Web Conference for the first time is walking the talk and really doing it through the internet,” says Yoelle Maarek, vice president for research and science in the Alexa Search organization and one of the 10 elected members of the International World Wide Web Conference Committee, which organizes The Web Conference. “The web community had been discussing this format for a while, but we had to have our arm twisted a bit, by drastic circumstances, to implement it. This is a paradox, since we, as the Web Conference, should have been the first one to embrace it. Hopefully, we will consider hybrid virtual/in-person formats going forward.”
Amazon researchers have eight papers accepted to the conference, with topics ranging from speech synthesis to knowledge graph management to product pricing on e-commerce sites. Amazon researchers — including Maarek, who will join a panel on conversational assistance — are also participating in six panels, talks, or workshops.
Among the major computer science conferences, Maarek says, the Web Conference is unusual in that it isn’t affiliated with a larger organization — an organizational structure that reflects the decentralized ethos of the web. “It kind of emerged from the internet and web community,” Maarek says.
As befits a conference that grew up around a single technology, The Web Conference is welcoming to applied research. “I think my first paper was on a bookmark organizer,” Maarek says. “Remember bookmarks? We built a plug-in to organize bookmarks automatically.” That attention to applications makes the conference particularly appealing to Amazon researchers, whose work is always customer focused.
“It’s not 100% applied — there are tons of theoretical works in the program,” Maarek says. “But yes, we want papers to be motivated by ‘considerations of usage’, as per Stokes’s ‘Pasteur’s quadrant’ model of basic and applied research.” One of the two papers at the conference from Maarek’s group, titled “Multi-Objective Ranking Optimization for Product Search Using Stochastic Label Aggregation”, is a case in point, Maarek says. “It is very theoretical, but it is motivated by a real customer problem. That’s typical of the Web Conference.”
The evolving web
Having attended the conference — which until 2018 was known as WWW — since the mid-’90s, and having been one of its program chairs, Maarek has been in a position to watch it evolve along with the web.
“It’s a big conference, with multiple tracks,” she says. “I remember when mobile and ubiquitous computing started to be very fashionable, and soon after, as expected, we had an associated track. Interestingly enough, though, mobile computing had a peak, and then it slowed down a bit. The track still exists but expanded to sensors and web of things. The next track that appeared and became very, very popular was about social networks, which attracted beautiful and elegant theoretical works with clear applications.”
In 2017, Maarek says, “the program committee chairs — one of them, Eugene Agichtein, is now an Amazon Scholar in my team — fought to introduce a new health research track, which seems particularly prescient this year.” Other recent additions are a track on crowdsourcing and human computation and one on the web and society. “Researchers started to understand the ethical implications of AI,” Maarek says, “and that with such technological power come responsibilities, too.”
Through all those shifts, however, “two things are a constant,” Maarek says. “Since its early days, the series focused on scale and on benefits to users. That makes it highly relevant to our Amazon scientists.”
Amazon papers at The Web Conference
“Collective Knowledge Graph Multi-type Entity Alignment”
Qi Zhu, Hao Wei, Bunyamin Sisman, Da Zheng, Christos Faloutsos, Xin Luna Dong, Jiawei Han
“Don’t Parse, Generate! A Sequence to Sequence Architecture for Task-Oriented Semantic Parsing”
Subendhu Rongali, Luca Soldaini, Emilio Monti, Wael Hamza
“Multi-objective Relevance Ranking via Constrained Optimization”
Michinari Momma, Ali Bagheri Garakani, Nanxun Ma, Yi Su
“Multi-Objective Ranking Optimization for Product Search Using Stochastic Label Aggregation”
David Carmel, Elad Haramaty, Arnon Lazerson, Liane Lewin-Eytan
“Study on Price Consistency regarding Pack Size via Product Variant Retrieval and Pack Size Extraction”
Yongning Wu, Chu Wang, Yang Liu, Zuohua Zhang
“Treating Cold Start in Product Search by Priors”
Parth Gupta, Tommaso Dreossi, Jan Bakus, Yu-Hsiang Lin, Vamsi Salaka
“Voice-based Reformulation of Community Answers”
Nachshon Cohen, Simone Filice, David Carmel
Panels, Tutorials, Workshops
Panel on Conversational Assistance | Yoelle Maarek
Fairness and Bias in Peer Review and other Sociotechnical Intelligent Systems | Nihar Shah and Zachary Lipton
Forecasting Big Time Series: Theory and Practice | Yuyang Wang, Christos Faloutsos, Valentin Flunkert, Jan Gasthaus and Tim Januschowski
Explainable AI in Industry: Practical Challenges and Lessons Learned | Krishna Gade, Sahin Geyik, Krishnaram Kenthapadi, Varun Mithal and Ankur Taly
Learning Graph Neural Networks with Deep Graph Library | George Karypis, Zheng Zhang, Da Zheng, Minjie Wang, Quan Gan
The Second Workshop on e-Commerce and NLP (ECNLP 2) | Shervin Malmasi, Eugene Agichtein (Amazon Scholar), Oleg Rokhlenko, Ido Guy and Nicola Ueffing
