SIGIR 2018 Review

The 41st International ACM SIGIR Conference on Research and Development in Information Retrieval took place at the University of Michigan in Ann Arbor July 8th – 12th 2018. As usual, the conference was highly competitive, with a 21% acceptance rate for long papers! For the first time, China surpassed the USA with most accepted papers.

Given the recent spotlight on privacy and ethical concerns in technology, I paid particular attention to what was presented around these topics. Tuesday morning provided one end of the spectrum, which kicked off with a keynote address on data science for social good and was followed by a session on methods to protect individual privacy in search. The other end of the spectrum was provided the following morning, with the location and trajectory session with methods that provide little to protect privacy and the mobile user behavior session (further discussed below) that had many concerns about digital ethics.

Aside from digital ethics and privacy concerns, there were many other sessions and events that took place over the course of 5 days.

Sunday

The opening welcome reception took place at the University of Michigan Museum of Art. This event was preceded by the tutorial sessions of which 11 different options were available. The doctoral consortium also took place on Sunday with 12 PhD students invited, including your correspondent, to present their ongoing research to a group of established researchers in the audience.

PhD Students and Mentors at the 2018 SIGIR Doctoral Consortium

Monday

The day kicked off with the presentation by soft spoken and humble Kalervo J√§rvelin, winner of the 2018 Gerard Salton award which is given out once per 3 years. He provided us with an overview of influential work for his research, such as work by Bates’ on interfaces. This award highlights the importance of including humans in our research and not just to focus on IR systems.

The remainder of the day included excellent sessions on log analysis and conversational systems, but most notably the day was centered around the students in IR. The student luncheon ran out of tickets, with over 200 attendees and mentors crammed into a packed room. Tetsuya Sakai provided an excellent comparison of academia and industry based on his experience in both domains. A student event in Bill’s Beer Garden took place in the evening to offer additional opportunities for students to converse with experienced researchers.


Student event at Bill’s Beer Garden in downtown Ann Arbor

Tuesday

Data Science for Social Good was the theme of the morning. The keynote speaker, Rayid Ghani, provided many examples of how his research group at the University of Chicago are tackling problems that will better the good of society. His presentation was a nice segue into the session for social good, of which fact checking methods for fake news as well as predicting human movement in emergencies were highly thought provoking topics. An RSVP only luncheon took place after these sessions to discuss concerns and approaches to include research from under-represented continents and countries.

Ice sculpture at the Henry Ford museum and conference banquet.

The conference banquet took place at the Henry Ford Museum. This was an exceptional banquet, in that it took place in a very unique location and also broke the tradition of sitting at tables while awards were presented. The museum provided attendees with many historical artifacts from the industrial revolution through present day, including the Kennedy presidential limousine and the Oscar Meyer Wiener Mobile. There was also a live Motown funk band that brought many of the attendees out onto the dance floor until we were given the boot.

Wednesday

Leif Azzopardi was scheduled to be the first speaker at the highly attended “New Metrics” session. He somehow managed to make a swap to be the last speaker, a wise choice given he was spotted on the dance floor at the end of the banquet dinner. This session introduced us to methods to measure utility of search results as well methods to consider user expectations.


In the mobile user session, we learned how mobile data can be used to determine who you talk to the most. Are there ethical concerns?

A later session in the morning included the focus of mobile user behavior introducing methods to identify user preferences. User preferences discussed included methods to identify if you prefer talking to your mother or your significant other. While this session provided plenty of focus on methods to understand preferences, there was very limited discussion on ethical concerns and implications of these methods.


Important parallel session to SIGIR business meeting.

The SIGIR Business meeting took place in the afternoon, however many attendees left this important meeting to view the equally important meeting between England and Croatia (see photo). All of us made it back in time to hear the panel and discussion regarding the future of information retrieval. One highly entertaining question and remark came from Norbert Fuhr regarding the lack of statistical analysis in many submissions this year, perhaps he has strong views around this topic?

Thursday

The workshops were an excellent way to round out the conference and also a signal to what tracks will be included at the 42nd SIGIR in Paris. The professional search workshop was what I found most interesting. In the domain of professional search, there are many problems that still need to be addressed, and Dave Lewis’ presentation on research challenges in IR really hit home on this. He pointed out that a part of speech tagger that is 98% accurate is still very inaccurate for sentences, paragraphs and documents, thus creating many problems downstream. Furthermore, in the domain of e-discovery, the two most significant advances are identifying duplicate documents and identification of blank pages. Professional search is clearly an open problem and provides a pathway for significant research.


Dave Lewis’ slide on 2 major advances in NLP for E-Discovery

For myself, this was my first time attending SIGIR and am very grateful for the ACM SIGIR student travel grant made available to make this experience possible. Also, many connections were made and discussions took place to better the research for myself and many hundreds of other attendees. I look forward to attending future SIGIR conferences.

About Steven Zimmerman
Steven Zimmerman

Steven Zimmerman is a PhD student at the University of Essex. His research interests are in the areas of IR, NLP and event prediction. He is currently involved with the Human Rights Big Data and Technology project. His work is focussed on development of methods to enable users with better tools and transparency for their experience in information retrieval platforms and to be better informed about the information they view. His research ultimately addresses concerns around freedom of expression, censorship, privacy as well as misinformation and toxic media.

Leave a Reply