Investigating Proactive Search Support in Conversations

Abstract

We are pleased to welcome Designing Interactive Systems DIS 2018 to Hong Kong. DIS 2018 is the first of its kind in two ways. It is in Asia for the first time. It is also in an art and design school for the first time. The theme of the conference is Design and Diversity. The theme reflects a classic design themeand also a foundational distinction in philosophy-of universals and particulars. Should we, as designers, follow Silicon Valley in its quest for products that engage everyone on the planet, or the architect Glenn Murcutt's conviction that he can only build in places he knows so well that his designs can be outstanding? The underlying logic of this question divides designers and design disciplines and emerges in every design process. This theme operated as a guiding tool for selecting our four keynotes, professors Jodi Forlizzi, Kun-pyo Lee, Phoebe Sengers and Erik Stolterman. During their years in design, they have lived through its diversities. We were happy that they accepted the challenge to share their experiences and thoughts about diversity to the benefit of our community. The theme was also our tool for directing the DIS community into the future. A few years from now, we hope, we will start to see answers to the challenges our keynotes are posing to us. The nucleus of the conference organization were two chairs and three technical chairs. This small group invited sixteen Subcommittee Chairs, three Pictorials chairs, and two chairs each for Workshops, Provocations and Work-in-Progress, Doctoral Consortium, and Demos. These chairs recruited 100 Associate Chairs, who recruited 1818 reviewers. Our review and decision schedule was brutal, but the organization worked through it efficiently, always with humor, and with collegial respect. DIS 2018 received 645 submissions: 405 for full papers and notes submissions; 71 for pictorials; 23 for workshops submissions; 107 for Provocations and Work-in-Progress (out of these, 19 were Provocations); 20 for Doctoral Consortium Submissions; and 19 for Demos. Acceptance rates were: 23% for papers and notes, 24% Pictorials, 55% for Workshops, 73% for Demos, 53% for PWiPs, and 50% for Doctoral Consortium. All this work led into a highly competitive conference between 9-13 June. June 9-10 were reserved for Workshops and Doctoral Consortium, and June 11-13 for 28 paper sessions. Pictorials are not in separate sessions; they are treated the same way as Full Papers. The 11th of June became an Experience Night of Demos, PWiPs, and a small design exhibition, which illuminated interaction design in Hong Kong and the Pearl River Delta. As extra, we organized a postconference trip to a few technology companies in Shenzhen, China. The program reflects the highly diverse nature of our discipline. We could not see any particular trends in papers, but a few observations may be possible. Two thirds of submissions focused on the following two human-oriented subcommittees: Experience and Methods and Processes. In comparison, the two technically oriented committees were smaller, but th...

'A Modern Up-To-Date Laptop' - Vagueness in Natural Language Queries for Product Search

“…To reach native English speaking participants, it is set up as an online study. We use the survey platform SoSci Survey 1 . For both the survey and the search task, participants are asked to use the internet browser on their private device.…”

“…As described by Lehtola, Heinecke and Bounsaythip [26], retailers and users use different vocabularies, likewise complicating the application of advanced natural language processing techniques used in conversational search, e.g. query expansion and the usage of word embeddings [1]. Various natural language processing methods have already been developed for learning distributed word representations [30,31] to address the vocabulary mismatch problem.…”

Naturalistic Digital Task Modeling for Personal Information Assistance via Continuous Screen Monitoring

“…However, the intent model required the users to provide explicit feedback and did not support proactivity feature. Our most recent work investigated a proactive search agent that could support natural spoken conversations by augmenting the conversations with additional information [1]. Both have shown the effectiveness of search systems in assisting users performing their tasks, but they were lab-based studies and tasks were designed to answer specific research questions.…”

“…The problem of task recognition has recently attracted a great deal attention throughout web search community [6,8]. Recognizing tasks is important to various competencies; it permits a system to determine user satisfaction with search results [11], predict future intents [9,10], and provide proactive assistance [7,1]. In traditional approaches, the main user signals used to recognize tasks typically rely on search engine query logs [12].…”

“…LSA learns a latent lower-dimensional representation of the input data. Each dimension in the lower-dimensional space can be interpreted and labeled as a representation 1 https://cloud.google.com/vision/ of a task, and used to retrieve task-relevant documents and labels.…”