Keywords

1 Introduction

Several idea creation methods have been proposed to help generating novel and useful ideas. Brainstorming method is widely used in companies, educations, communities, etc. [1, 2] because of the facile introduction.

KJ-method is one of the most popular method used in brainstormings [3]. In KJ-method, each idea submitted by the participants is written on a small card. By arranging the cards, participants can find relationships between submitted ideas during the discussion, and it helps them to submit more ideas and conclude the discussion.

Although KJ-method is a powerful tool, several weak points also have been pointed out. For instance, participants have to submit ideas as much as they can, but generating much new ideas is not easy for ordinal participants. In brainstormings, submitting novel ideas is required for the participants. However, it is also hard for ordinal participants because they tend to get effect from ideas which already submitted in the discussion.

In this paper, a new supporting system for brainstorming has been proposed in order to solve these problems. The proposed system shows participants some related images of important keyword used in the discussion. Participants can create novel ideas by referring the images.

2 Related Works

Several methods for supporting brainstorming processes using digital devices have already proposed.

Miura et al. have proposed a system with a transmitted light table for supporting to conduct brainstormings [4]. By using the system, users can save processes and results of brainstormings easily. Aikawa et al. have proposed a system in which combined tablet devices are used [5]. With the system, users can conduct brainstormings anywhere together. There systems focus to support to conduct brainstormings, and supporting to generate new ideas is not the scopes.

Kang et al. have developed Category Camera for supporting to create new idea in brainstormings [6, 7]. This mobile application was developed aimed to support fieldwork step of group works. The users can take photographs and put labels on them easily by using this mobile application. The labels help users to classify the photographs in discussion phases.

Matsuo et al. have developed a supporting system for idea generation [8]. First, a user input a topic word and a keyword which he/she thinks important for the topic. Then the system shows some words which relate to the topic word and the inputed keyword. Next, user selects several words from the shown words that he/she decides important for the topic, and then the topic word, the related words, and the selected words are connected together via a word which explains the relationships. In this way, the users can marshal relationships between ideas and the topic.

In comparison with these previous systems, the system proposed in this paper aims more instinctive ways. The proposed system represents some related images of submitted idea for users in order to help them to generate diverse and novel ideas.

3 Outline of the Proposed System

The aims of the proposed system is to solve the following two problems during brainstormings:

  • It is not easy to submit many ideas in a limited time.

  • It is difficult to submit novel ideas when we are influenced by preconceived ideas and images.

The proposed system shows the users related images of the important keywords used in ideas submitted in discussions. Figure 1 shows the outline of the proposed system.

Fig. 1.
figure 1

Outline of the proposed system

Full size image

 

(1):

Participants submit their idea by using their own smartphones. The submitted ideas are stored in the DB on the Internet.

(2):

Characteristic words are extracted from the submitted ideas.

(3):

Related words of the extracted words are generated by using an external service.

(4):

Related images are obtained by retrieving the related words as the retrieval keywords on the Internet.

(5):

Pairs of idea and related images are presented for the participants.

 

The system shows hints of discussions as images instead of words. Hence, the users can get the hints instinctually without interrupting thinkings, and they can focus on the topics.

The system has another remarkable point in step (3) and (4) in Fig. 1. The extracting characteristic words are not used for obtaining related image directly. Related words of the characteristic words are generating first, and then the related images are obtained by using these related words as keywords. Because of this mechanism, obtaining unpredictable images is expected. These unpredictable images can help to vitalize discussions, and also help to escape form stagnating discussions.

4 Experiments

Experiments have been conducted in order to investigate the effects of related images on discussions. Numbers of utterances in brainstormings were measured to evaluate the activation levels.

The participants were 8 undergraduate students of engineering course in a Japanese national university. 4 of them were 3rd grade and the others were 4th grade. The participants were divided into 2 groups, Group A and Group B. They were asked to conduct brainstormings under two conditions, with related images and without related images.

In this experiments, kizAPI [9] was used for generating related words, and Bing search API [10] was used for obtaining related images.

4.1 Process of Experiments

A preliminary experiment was conducted beforehand. The purposes of the preliminary experiment are checking the participants’ skill of brainstorming and accustoming the participants to the system. In the preliminary experiment, each group was asked to conduct a brainstorming with the proposed system without showing related images for 5 min. The topic was decided as “ways to continue diet” by considering the difficulty.

During the preliminary experiment, the number of submitted ideas was counted. As a result, the both of the group submitted the same number of ideas, 11. Thus it can be assumed that the two groups have the same level of skills of brainstorming.

The flow of the experiments is shown below:

  1. 1.

    Instructions of the proposed system and the experiments

  2. 2.

    Tutorial and exercise of the system

  3. 3.

    Break (3 min)

  4. 4.

    Brainstorming with topic 1 (10 min)

  5. 5.

    Break (3 min)

  6. 6.

    Brainstorming with topic 2 (10 min)

The following two topics were used in the experiments.  

Topic 1. :

Ways to reduce smoking while walking

Topic 2. :

Ways to vitalize a life on Monday

 

Table 1 shows the combinations of topics and conditions. In order to reduce the effect of oder, Group A conducted under condition 1 first, meanwhile Group B conducted under condition 2 first.

Table 1. Conditions of the experiments (Condition 1: with related images, Condition 2: without related images
Full size table

4.2 Results and Discussions

Number of Ideas. Table 2 shows the number of ideas submitted during the experiment under each condition.

Table 2. The number of submitted ideas (Condition 1: with related images, Condition 2: without related images
Full size table

As we can see from the table, there is no difference between the number of utterances under condition 1 and under condition 2 for group A. Group A conducted brainstorming under condition 1 (with related images) first, and then conducted under condition 2 (without related images). Thus, the adjustment to the task might effect on the results. Meanwhile, Group B, they conducted brainstorming under condition 2 first. As a result, they submitted about 1.5 times more number of ideas under condition 1 than under condition 2. The fact also supports the assumption of the effects of adjustment.

The difference of difficulty of the topics also could effect on the results. For the topic 1, both of the group tended to submit common-sense ideas. On the other hand, both of the group could submit diverse ideas for the topic 2. Group A discussed topic 2 under condition 2. The facts could affect on the results.

Fig. 2.
figure 2

Number of utterances during discussions

Full size image
Table 3. Utterances list
Full size table

Flow of Discussions. Table 3 shows all of the utterances of group A for each 10 seconds under each condition. The left side shows the utterances under condition 1, and right side shows the utterances under condition 2. The columns labeled “Number of utterances” show number of utterances which brought out in the period. An ID number is put for each utterance.

Yellow highlighted columns show the ideas which might be inspired by related images provided by the system. For instance, I057: “Let all fortune information in morning television show make good luck” was submitted by refering an image of guide of palmistry. I070: “Have deluxe breakfast in each Monday” was submitted by refering an image of a huge hamburger. These idea could not be expected without the images.

The graph shown in Fig. 2 represents the transition of number of utterances of every ten seconds of group A. The solid line shows the number under condition 1 (with related images) and the dashed line shows the number under condition 2 (without related images).

Generally, lively discussion is hardly expected in the beginning of a discussion because the atmosphere is not activated yet. In accordance with the fact, the number of utterances was not stable in the beginning (from 00:00 to 01:30, Fig. 2-A) under condition 2 (average:3.33, SD:1.94). On the other hand, the number of utterances was stable in the same period under condition 1 (average:4.11, SD:1.63).

In the closing period (from 08:30 to 10:00, Fig. 2-B), the number of utterances was decreasing under condition 2 (average:1.90, SD:1.97) while active discussion was continuing by being inspired by related images under condition 1 (average:4.20, SD:1.87).

For the middle period (from 01:30 to 08:30), there are not significant difference between under condition 1 (average:3.79, SD:1.18) and under condition 2 (average:3.86, SD:1.20). Generally, lively discussions can be expected in middle period of discussions. Thus, they could keep active discussions despite with the supporting system.

5 Conclusion

In this paper, a supporting system for brainstorming has been proposed. The system aims to help users generating novel and divers ideas by showing related images of ideas submitted in discussions.

Experiments were conducted in order to investigate the effects of showing related images on generating new ideas. As a result, it has been found that discussions in beginning period and closing period are vitalized by related images.

Unfortunately, the differences are not shown statistically. Additional experiments under strict conditions should be scheduled as a future work. In addition, implementation of the system with user friendly interface is also required. The internal design and interface design have already completed [11], and the implementation is progressing.