Continuous Display of Images Searched by Keywords Extracted from a Brainstorming Session and Suggested by an Autosuggest Function Daisuke Shibata Yutaka Yamaguchi Graduate School of Graduate School of Science and Engineering, Science and Engineering, Saga University. Saga University. 18573007@edu.cc.saga-u.ac.jp 18634011@edu.cc.saga-u.ac.jp Chika Oshima Koichi Nakayama Faculty of Science Faculty of Science and Engineering, and Engineering, Saga University. Saga University. chika-o@ip.is.saga-u.ac.jp knakayama@is.saga-u.ac.jp ABSTRACT In this paper, we presented a system called SWISS (Search Websites’ Images using Search Sugges- tions) that searches images for brainstorming activ- ities. Participants’ utterances are continuously in- put into SWISS and converted into text. SWISS searches Webpages using the text and extracts four words, which are features of the top Webpage of the search results. These words are expected to be a tool for deepening the current conversation topic. Moreover, each word of the four words is listed with a letter of the alphabet/Hiragana char- acter, which is selected at random and receives a suggested word. Then, SWISS searches images with extracted and suggested words. We conducted an experiment that three people discussed given themes with using SWISS and using SWISS with- out autosuggested word. The result showed that a participant could come up with a new idea from un- expected images displayed from SWISS. KEYWORDS Idea generation support system, Speech recogni- tion, Image retrieval, IBM Watson, Autocomplete function. 1 INTRODUCTION There are many methods/systems that encour- age people to generate and/or crystallize an idea [1][2][3][4][5][6]. Companies usually want to be successful at developing attractive products that incorporate novel ideas. When the ideas are generated, using typical informa- tion will not help to achieve the goal of novelty [7]. Some companies have an online open idea generation platform by which consumers can propose ideas to the company [8]. In this way, diverse opinions lead to an increased breadth of idea production [9]. Wang [10] stated, “Using pictures as extra stimuli may be more effective than language in stimulating idea generation.” Thus, “Idea- Expander” [11] is a tool that supports group brainstorming by displaying pictures based on a chat conversation. IdeaExpander monitors the chat conversation and determines whether a remark contains an idea by using a machine learning classifier [10]. Then, the keywords drawn from the remark are used to search pic- tures on Flickr. Especially, “pictures that con- tain multiple idea categories or categories that are less commonly discussed” help to convert the diversity of generated ideas [10]. There- fore, when people are brainstorming, a picture that illustrates not only the meaning of word “X”, which they have already put forward, but also the meaning of another word, “Y”, with which they have not yet associated the word “X”, may encourage their idea generation. In this paper, we present a system, SWISS (Search Websites’ Images using Search Sug- gestions), which continuously extracts words from the utterances of participants in a discus- sion. Then, the system constantly displays pic- ISBN: 978-1-941968-52-9 ©2018 SDIWC 76 Proceedings of The Fourth International Conference on Electronics and Software Science (ICESS2018), Japan, 2018