Idea discovery: A scenario-based systematic approach for decision making in market innovation

https://doi.org/10.1016/j.eswa.2012.07.044Get rights and content

Abstract

A new trend of researches on knowledge discovery and chance discovery is to identify human insights through data synthesis rather than to discover facts through data analysis. In this paper, we propose a systematic approach named idea discovery which is committed to turning data into effective human insights. Idea discovery focuses on dynamic and sustainable process for high-quality ideas cultivation, construction, integration and evaluation through human–computer and human–human interaction. It mainly relies on latent information and its dynamic changes to drive ideas creation, integration and evaluation during sustainable creativity process. The process of idea discovery is in accordance with a dynamic model which contains two key components: (1) mining algorithms to turn data into scenario maps for eliciting human insights; (2) scenario-based creativity support activities towards actionable ideas generation. An intelligence system called Galaxy integrated with IdeaGraph algorithm has been developed to support the dynamic process of idea discovery. A case study in an automobile company has validated the effectiveness of proposed method and system.

Highlights

► We propose a systematic approach named idea discovery to identify human insights through data synthesis. ► Idea discovery focuses on uncovering latent information and its dynamic changes for ideas creation and evaluation. ► We have developed an intelligence system called Galaxy to support idea discovery process. ► A case study in one of world famous auto companies has verified the effectiveness of proposed method and system.

Introduction

Text-based knowledge discovery is increasingly playing an important role in scientific discovery. It is helpful for researchers to manage valuable information and identify facts in text data that would solve practical problems. On the other hand, early researches on literature-based discovery (LBD) are focused on the scientific field of biomedicine. Swanson (1986) made a new discovery in text literatures and proposed a hypothesis that fish oil could cure Raynaud’s disease. Finally, he tested the hypothesis experimentally and clinically (Swanson, 1986, Swanson, 1987, Swanson, 1988, Swanson, 1989). To solve the complexity of text processing, information retrieval (IR) and information extraction (IE) are used to develop system tools to support discovery process (Gordon and Lindsay, 1996, Lindsay and Gordon, 1999, Swanson and Smalheiser, 1997, Swanson and Smalheiser, 1998, Weeber et al., 2001). For text-based understanding, information extraction potentially offers better support than information retrieval as it can extract useful information and relations expressed in the text documents. It is a good way to interpret extracted information by visualizing terms and relationships graphically (Mack & Hehenberger, 2002).

Text-based knowledge discovery is also widely applied in other areas. For instance, criminal investigators may make use of text data to uncover evidence. A general framework is proposed by combining information extraction techniques with visual exploration techniques to provide an approach to make evidence discovery (Chen et al., 2004, Louris and Engelbrecht, 2011). In business, patent-based discovery is used by enterprises or stakeholders to timely be aware of the situation and direction of current technologies so as to rapidly adjust their market strategies (Chen et al., 2004, Fattori et al., 2003, Losiewicz et al., 2000).

Chance Discovery, proposed by Ohsawa in 2000, is a relatively new research field as an extension of text-based knowledge discovery. It is a human–computer interaction process to detect rare but important chances for decision making. A chance means to understand an unnoticed event or situation which might be uncertain but significant for a decision (Ohsawa & McBurney, 2003). A core visualization tool called KeyGraph can generate scenario map to aid human’s value cognition the double-helix process of chance discovery. In fact, KeyGraph is a keyword extraction algorithm from a single document using co-occurrence graph (Ohsawa, Benson, & Yachida, 1998). That is, a document is represented as a graph where each node corresponds to a term and each edge means the co-occurrence of two terms. Based on the segmentation of a graph into clusters, KeyGraph extracts keywords by selecting the terms which strongly co-occurs with clusters. Fig. 1 shows a scenario map visualized by KeyGraph. The red1 nodes are considered as chance candidates because they act as a bridge linking different sub-scenarios. Human are required to understand and interpret the importance of chance candidates that can make the situation transfer from one sub-scenario to another. Later, Ohsawa (2005) proposed a breaking-through method named data crystallization where dummy nodes representing invisible events are inserted into the processed text data, and then new data is visualized by KeyGraph. However, the complex algorithm and graph obtained were hard for users to understand, thus Maeno and Ohsawa (2007) subsequently present a new method, human–computer interactive annealing, for revealing latent structures and discovering dark events. Based on chance discovery, Hong (2009) proposes an interactive human–computer process model called Qualitative Chance Discovery Model (QCDM) to extract more accurate data representation in scenario map for decision making on potential chances.

In recent years, a scenario map generated by KeyGraph with data crystallization has been applied as the game board in Innovators Market Game ® (IMG) and Innovators Marketplace ® (IM) for innovative chance discovery (Ohsawa, 2009, Ohsawa et al., 2010, Ohsawa and Nishihara, 2012). In particular, human cognition of KeyGaph scenario map has been expanded from sensing bridge-like chances to creating chances through combining black nodes, see Fig. 2. Wang and Ohsawa, 2012a, Wang and Ohsawa, 2012b have proposed a systematic approach with IMG for ideas creation and evaluation, and have applied such an approach in product design. To improve IMG, a customer-centric creativity support techique, 4W-IMG, has been proposed (Wang & Ohsawa, 2011).

Although many relevant researches have successfully achieved knowledge discovery and business chance discovery for decision making, a new trend in discovery researches is to identify true human insights through data synthesis rather than data analysis.

In this research, we propose a novel systematic approach named idea discovery which extends static discovery into dynamic discovery in three dimensions. A hypothesis is proposed that latent information would be captured across different dimensions. Section 2 summarizes current discovery pattern and introduces the process of dynamic discovery. Idea discovery with a dynamic model is presented in Section 3. In Section 4, a mining algorithm called IdeaGraph as a key component of idea discovery is presented for discovering more latent information (event and their relations). A case study in automobile industry is described in Section 5 and the conclusion is in Section 6.

Section snippets

Discovery pattern and dynamic discovery

Previous researches on text-based knowledge discovery and chance discovery both combine two complementary processes. One is the process of Computer’s text data mining and visualization; the other is the process of human’s cognition, understanding, interpretation and innovation. Fig. 3 shows a basic framework of current discovery pattern.

Dynamic discovery is commonly used by enterprises to evaluate and understand technology trends through patent data analysis, and eventually achieve a strategic

Idea discovery

Idea discovery, as an extension and evolution of previous discovery researches, is a dynamic and sustainable process for high-quality ideas cultivation, construction, generation and evaluation through human–computer and human–human interaction. Idea discovery not only works on rare and important event points, but also focuses on latent and significant event relationship and the dynamic changes of these events and their relationship. Therefore, idea discovery is committed to digging up latent

IdeaGraph: an algorithm of exploiting more latent information for eliciting more human insights

In this section, we introduce a human-oriented algorithm called IdeaGraph which can generate a rich scenario graph for human’s comprehension, interpretation and innovation. IdeaGraph not only works on discovering more rare and significant chance events, but also focuses on uncovering latent relationship among them. Suppose that data has been preprocessed into D.D=item1,item2,item3,item4item2,item7,item5item3,item6,item9,item10,item5Fig. 6 shows a scenario map formation process by

An overview of Galaxy

An intelligence system named Galaxy has been developed to support the process of idea discovery based on a dynamic discovery model shown in Fig. 5. IdeaGraph has been integrated into the Galaxy system. Besides, Galaxy employs an automatic graph layout algorithm to visualize directed and undirected graphs (Fruchterman & Reingold, 1991).

Fig. 7 shows a screen shot of Galaxy. There are five main modules in Galaxy: Solution Explorer, Graph Viewer, Metadata Display, Data Dialog Box and Dynamic

A case study in automotive industry

We have successfully carried out a project in a famous auto company. The objective of the project is to explore Chinese users’ preference on human–machine interface (HMI) system for further development. HMI system may aid users’ simple access to all infotainment components, such as navigation, telephone, video/audio, entertainment information, etc. To improve user experience while driving, the company is eager to discover more potential business opportunities which help to create new ideas and

Conclusion

Traditional text-based discovery researches too much focus on exploring the facts through data analysis. However, a new trend is to identify true human insights through data synthesis rather than data analysis.

In this paper, we examined a scenairo-based system approach called idea discovery for sustainable creativity support through human–computer and human–human interaction. Unlike traditional text-based discovery methods, idea discovery mainly depends on uncovering latent and important

Acknowledgments

The author was supported through the Global COE Program “Global Center of Excellence for Mechanical Systems Innovation”, by the Ministry of Education, Culture, Sport, Science and Technology, Japan, and was also funded by Chinese Scholarship Council.

References (28)

  • P. Losiewicz et al.

    Textual data mining to support science and technology management

    Journal of Intelligent Information Systems

    (2000)
  • A.L. Louris et al.

    Unsupervised discovery of relations for analysis of textual data

    Digital Investigation

    (2011)
  • Y. Maeno et al.

    Human-computer interactive annealing for discovering invisible dark events

    IEEE Transactions on Industrial Electronics

    (2007)
  • Y. Ohsawa

    Data crystallization: chance discovery extended for dealing with unobservable events

    New Mathematics and Natural Science

    (2005)
  • Cited by (33)

    • A literature review on individual creativity support systems

      2017, Computers in Human Behavior
      Citation Excerpt :

      A system called combinFormation supports information collection, integration, and visualization and is shown to improve idea generation (Koh, Kerne, & Hill, 2007). Similarly, providing a knowledge base or case-based reasoning system (Althuizen & Wierenga, 2014; Cheung, Chau, & Au, 2008; Forgionne & Newman, 2007; Liu et al., 2011), visualizing information or the concept space (Brade, Heseler, & Groh, 2011; Fentem, Dumas, & McDonnell, 1998; Hori, 1997; Sugimoto, Hori, & Ohsuga, 1996), using data synthesis for uncovering important information (Wang & Ohsawa, 2013), and organizing research notes (Aihara & Hori, 1998) have all been used to support information finding. Jenkin, Chan, Skillicorn, and Rogers (2013) developed a system to identify novel information related to a topic.

    • Early detection method for emerging topics based on dynamic bayesian networks in micro-blogging networks

      2016, Expert Systems with Applications
      Citation Excerpt :

      Ohsawa, Benson, & Yachida (1998) used a KeyGraph algorithm to convert terms into a graph based on co-occurrence relations, and employed a community detection approach to partition the graph. Wang & Ohsawa (2013) proposed a systematic approach to turning data into effective human insights. Chen & Neill (2014) considered all potentially useful information in a unified nonparametric statistical framework, and facilitated the early detection of social events.

    • Creativity support systems: A systematic mapping study

      2016, Thinking Skills and Creativity
    • A hybrid term-term relations analysis approach for topic detection

      2016, Knowledge-Based Systems
      Citation Excerpt :

      How to discover important rare topics? In Chance Discovery (CD) theory and Idea Discovery (ID) theory, a chance is defined as a rare but important event or situation which has a strong impact on human decision making [9–11]. CD and ID as extensions of Knowledge Discovery have been used to detect chances by uncovering latent co-occurrence relations among terms.

    View all citing articles on Scopus
    View full text