2023-05-23: ACM SIGIR Conference on Human Information Interaction and Retrieval (CHIIR) Trip Report

By -


The eighth ACM SIGIR Conference on Human Information Interaction and Retrieval (#CHIIR2023) was held online in Austin, Texas, from March 19 to 23. After three years of virtual conferences, CHIIR 2023 switched to an in-person and online hybrid format. We (Bhanuka, Mohan, Dr. Vikas Ashok, and Dr. Sampath Jayarathna) presented our demonstration titled "DisETrac: Distributed Eye-Tracking for Online Collaboration" in person in Texas.

CHIIR is a well-established multi-and interdisciplinary conference with a research community engaged in human-computer interaction, information retrieval, information behavior, and human-information interaction. The full and perspective paper submissions of CHIIR 2023 comprised 26 accepted publications with an acceptance rate of 39.4%, presented over three days in nine sessions. There were 20 short papers (acceptance rate: 37.7%) and eight demonstrations (acceptance rate: 61.5%). Both demonstrations and short papers were presented in a joint session as posters at the conference venue, both in-person and online, on the first day of the conference. The conference also had ten doctoral consortium publications presented on the last day of the conference. 

Day 1: 

The first day of the conference had a keynote session, three paper sessions, and a joint session for short paper and demonstration submissions. The first day of the conference started with the keynote "Veracity, viewpoints, and vacancies: stories of knowledge gathering, compilation, and sharing" by Dr. Elizabeth Churchill,  senior director of UX at Google. The keynote presented interesting facts on search, information seeking, and exploration, with the potential future of information exploration combined with the recent advancements in AI, such as the effects of novel generative large language models such as ChatGPT. In addition, the conference also had an informal session to share ideas among community to further discuss the topic following the keynote sesssion. 


The first paper session of the conference included papers on task-oriented information retrieval. The first paper of the session was the perspective paper titled "Taking Search to Task." The study focused on the widely accepted assumption that searching is a collection of queries instead of a user attempt to accomplish a task. During the presentation, the authors established the importance of identifying, understanding, and addressing the users' tasks considering the recent developments in conversational agents. The subsequent papers of the session, titled "Understanding Recruiters' Information Seeking Behavior in Talent Search" and "Understanding Procedural Search Tasks in the Wild," presented studies on information-seeking behavior among recruiters and the general public, respectively.

The second paper session included papers on the design of information retrieval systems. The first paper, "Guiding Oral Conversations: How to Nudge Users Towards Asking Questions?" was about designing a conversational approach that nudges the information seeker (user) toward unrestrictive and immersive interactions. The session also included a study on potential immersive search engine result page (SERP) methods in virtual reality and their effects on user behavior using eye tracking, titled "Eyes on Immersive Search: Eye-Tracking Study of Search Engine Result Pages in Immersive Virtual Environments." The study provided valuable insights into presenting SERP results to users in head-mounted displays, including the positional biases and their perception of the effort on information acquisition. The session's final presentation ("The Evolution of Web Search User Interfaces - An Archaeological Analysis of Google Search Engine Result Pages") was on the evolution of Google's web search user interface using web archives. The study analyzes the evolution of SERP in terms of content, layout, design navigation, and file size while portraying the evolutionary trends over time.

The final session for the day was on obstacles to information retrieval. The first presentation was on the retraction of scientific articles and the prevention of their continued usage (Rising of Retracted Research Works and Challenges in Information Systems: Need New Features for Information Retrieval and Interactions). The study highlighted the increased usage of article retractions in various domains and proposed a framework to identify retractions and mitigate their use in post-retraction. The second paper was "Driven to Distraction: Examining the Influence of Distractors on Search Behaviours, Performance and Experience," a study that analyses the effects of different types of distractions on web pages by analyzing their satisfaction, workload, topic recall, and ease of concentration. Based on their results, the authors ranked the effects of different types of distractors on the user experience and the outcomes to magnify depending on the user's cognitive abilities. The study received an honorable mention for its contribution. The final paper of the day was on predicting the music skipping of the users using reinforcement learning, titled "Why People Skip Music? On Predicting Music Skips using Deep Reinforcement Learning".

 

Day 2:

The second day of the conference started with the keynote by Dr. J Stephen Downie titled "Why We Retrieve: An (Im)precise Recalling of 30 Years Leading from Behind". In the keynote, Dr. Downie discussed the evolution of music information retrieval over time. Then he shared his experience at Hathi Trust Research Center (HTRC) working on information retrieval projects, including working with a workset of over 6 billion pages highlighting similarities with Music Information Retrieval. The keynote ended with him sharing his suggestions for information retrieval for the future.
 

The first paper session (session 4) of day two contained studies on information retrieval and user disorientation. In the first study, titled "True or false? Cognitive load when reading COVID-19 news headlines: an eye-tracking study", authors have investigated the cognitive load imposed by reading COVID-19-related claims. For the study, the authors have used pupillary eye-tracking data. In the following study titled "It is an online platform and not the real world, I don't care much: Investigating Twitter Profile Credibility With an Online Machine Learning-Based Tool," the authors have developed a real-time tool to assess the credibility of a Twitter profile. Their findings reveal that the latest tweets and the behavior of retweets are the most contributing factors to the profile's credibility.

The second paper session (session 5) comprised information retrieval and academic work studies. The session included three papers investigating the information retrieval patterns of the users 1. Research Data Archives (Direct, Orienting, and Scenic Paths: How Users Navigate Search in a Research Data Archive), 2. Scholarly literature (How Data Scientists Review the Scholarly Literature), and 3. Autonomous learning (Incubation and Verification Processes in Information Seeking: A Case Study in the Context of Autonomous Learning).

The final paper session (session 6) started with "SearchIdea: An Idea Generation Tool to Support Creativity in Academic Search," a study on developing an enhanced search engine results interface to improve the brainstorming process. The tool allows users to save, organize, and brainstorm on search results while engaging in an information retrieval task. The authors had identified their approach to produce more favorable outcomes than the conventional approach of the select-save to a text pad. The presentation won the best paper award in CHIIR 2023. 

The next presentation was "The Infinite Index: Information Retrieval on Generative Text-To-Image Models," which presented the idea of using prompt engineering to develop an interactive text-based image retrieval system. The final presentation of the session, "One of Us: a Multiplayer Web-based Game for Digital Evidence Acquisition of Scripts through Crowdsourcing," presented an "Among Us" styled game where players identify digital traces left by each other.

Day 3:

The conference's final day included three paper sessions followed by the "CHIIR-It-Forward," a discussion session for the CHIIR community. The first session (session 7) had presentations on the influence of information retrieval on user behaviors. In the first presentation, titled "Consumer Health Information Quality, Credibility, and Trust: An Analysis of Definitions, Measures, and Conceptual Dimensions," the authors have reviewed literation on the concepts of credibility, trust, and quality and have identified them to share a core set of dimensions. In the second presentation "Investigating the Influence of Featured Snippets on User Attitudes," the authors presented their results on the influence of the featured snippets on user attitudes. The final paper of the session presented a study of using modern stance detection approaches to assist users in navigating search results on debated topics.

The day's second session had papers on machine learning models and their interpretability. The study titled "Understanding the Cognitive Influences of Interpretability Features on How Users Scrutinize Machine-Predicted Categories" provided exciting insights into the effects of tools and tools and visualizations for integrating machine learning interpretations in user interfaces. 

The final paper session of the day on elucidation covered papers on broader topics. The first paper of the session ("Much Ado About Gender: Current Practices and Future Recommendations for Appropriate Gender-Aware Information Access") was on the study of achieving gender fairness during research studies and in the development of systems through a systematic review of articles on information retrieval and recommender systems. The paper proposes recommendations for researchers and developers for well-grounded use of gender. The next session paper was "Building a Better Mousetrap: Tools and Processes for Selling A Company," a study on document management in early-stage startups and on developing a potential solution. The conference's final paper was "An Instrument for measuring users' meta-intents," where the authors proposed the concept of meta-intents to represent higher-level user preferences on interaction and decision-making and found them to be independent of the demography of the user.

CHIIR-it-Forward

With the end of all paper sessions, the conference's closing ceremony, "CHIIR-it-Forward," started with a presentation by Dr. Dana McKay on the origin of CHIIR by merging ACM IIIX and ACM HCIR followed by the evolution of CHIIR during the past years. The session included updates on the steering committee members, such as the appointment of new members and the new chair (Dr. Dana McKay) and the retirement of members. The session also included details on the upcoming CHIIR 2024 in Sheffield, England, and the plan for the conference, followed by the announcement of CHIIR 2025 in Melbourne, Australia. The session ended with a public discussion among the community on the suggestions for the future of the CHIIR community.

-- Bhanuka (@mahanama94)

Comments

Post a Comment