Last week we held the HCIR 2012 Symposium in Cambridge, Mass. This is the sixth in a series that we have organized. We expanded the format of this year’s meeting to a day and a half, and in addition to the posters, search challenge reports, and short talks, we introduced full papers reviewed to first-tier conference standards. I will write more about these later, and for details on other events at the Symposium, I refer you to the excellent blog post by one of the other co-orgranizers, Daniel Tunkelang.
In this post, I wanted to record my impressions of the keynote talk by Marti Hearst from UC Berkeley.
Marti Hearst delivered an interesting and entertaining keynote address titled “Halloween Cauldron of Ideas for Research (HCIR)” in which she tried to outline some potential directions for future HCIR research. Her talk was divided into three main themes: the seams of sense-making, multiple cognitive channels, and radical collaboration, and included several periods of discussion with the audience.
Seams of sense-making
With respect to sense-making, Marti identified saving search results and how to stay in the “flow” as being under-represented in the existing literature and systems. Saving search results is an important aspect of exploratory, recall-oriented search.
She illustrated some of the design challenges with an example from an existing information retrieval system that supports this functionality, but does not offer a smooth interface for doing so. This aspect was interesting to me as some of the work we’ve done on Querium sought to address issues of saving documents while remaining in the flow.
The notion of “flow” (Csikszentmihalyi, 1975) that characterizes engaged, uninterrupted, intrinsically-motivated behavior is also important for information seeking. Marti talked about the need to reduce friction of interaction, and the need to integrate note-taking into exploration. She also mentioned some of the advantages of spatial hypertext for organizing found information, but wanted to see a more light-weight interaction that did not detract from the flow.
During discussion, several other challenges were raised including how to keep track of new query ideas, how to implement transclusion (Nelson, 1999), and how to manage interactions with standing queries and alerts.
In addition to text, Marti suggested that designers need to think about other modalities such as audio to augment certain kinds of interactions with information seeking systems. Integrated audio interaction is something of a missed opportunity because it is difficult to work with in current programming frameworks. Yet there are advantages to audio both as input and as output: Marti cited work by Soudian and Fels (2002) that showed that programmers are more likely to leave detailed and informative comments in audio than in text; audio interfaces for textbooks, designed specifically for people with visual impairments, have also been shown to be useful to people with normal sight. Despite some advantages of audio, significant challenges exist in integrating it into interfaces in cognitively-appropriate ways.
Marti’s third theme was titled “radical collaboration.” She described several variants on collaborative search, which she classified into four categories: “My results affect your results (and vice versa),” “my search is independent of yours (until I am ready to share),” “Ill search more if I see you are (even if you are a fakebot),” and “I’ll search with 100 other people (crowdsourcing complex finding tasks).” These scenarios reflect some of the interesting and significant use cases for collaboration in information seeking, an important emerging area of HCIR.
This aspect of the talk was clearly influenced by some of the ideas on collaborative search developed at FXPAL, and one of the images Marti showed was a screenshot of two participants in a Cerchiamo study of collaborative search.
In her presentation, Marti raised some interesting questions and challenges, some of which we will surely see discussed in next year’s meeting and in publications in related conferences such as CHI and SIGIR. It was also pleasing to me personally to see a considerable degree of agreement between Marti’s vision of the HCIR future and my own research agenda.
Csikszentmihalyi, M. (1975). Beyond boredom and anxiety. San Francisco: Jossey-Bass.
Nelson, T.H. (1999) Xanalogical structure, needed now more than ever: parallel documents, deep links to content, deep versioning, and deep re-use. ACM Computing Surveys, 31(4), December 1999.
Soudian, S., Fels, D.I. (2002). Verbal source code descriptor. Proceedings of IEEE WESS’ 2002.