Previous studies of innovation have recognized that many innovations are developed by users. However, there is a risk of leaking new ideas by users who join a discussion to generate ideas. In order to avoid the risk, this study proposes a new workshop method to generate business ideas. In the workshop method, idea generators are required to discuss new business ideas based on information that is organized by users who do not join the discussion and thus never know new ideas that are created in this workshop. Idea generators who are given the user-organized information are considered to be able to create new ideas using the given information. We conducted an experiment to test this. In our experiment, participants were divided into two groups: the first group was asked to generate new business ideas based on the information with user perspective while the second group was asked to do so based on the information with engineer perspective. Performance of the first group was compared with that of the second group. Eight outside experts rated all ideas generated in terms of novelty, benefit and feasibility. The result showed that the ideas generated by the first group were rated significantly higher in terms of novelty as well as lower in terms of feasibility than the ideas generated by the second group. Furthermore, a questionnaire survey carried out to those who joined this workshop supported this finding. Our findings suggest that our workshop method is useful for bringing user perspective into actual business idea generation.
Recently, natural language processing research has begun to pay attention to second language learning. However, it is not easy to acquire a large-scale learners' corpus, which is important for a research for second language learning by natural language processing. We present an attempt to extract a large-scale Japanese learners' corpus from the revision log of a language learning social network service.This corpus is easy to obtain in large-scale, covers a wide variety of topics and styles, and can be a great source of knowledge for both language learners and instructors. We also demonstrate that the extracted learners' corpus of Japanese as a second language can be used as training data for learners' error correction using a statistical machine translation approach.We evaluate different granularities of tokenization to alleviate the problem of word segmentation errors caused by erroneous input from language learners.We propose a character-based SMT approach to alleviate the problem of erroneous input from language learners.Experimental results show that the character-based model outperforms the word-based model when corpus size is small and test data is written by the learners whose L1 is English.
Multi-document summarization is the task of generating a summary from multiple documents, and the generated summary is expected to contain much of the information contained in the original documents. Previous work tries to realize this by (i) formulating the task as the combinatorial optimization problem of simultaneously maximizing relevance and minimizing redundancy, or (ii) formulating the task as a graph-cut problem. This paper improves summary quality by combining these two approaches into a synthesized optimization problem that is formulated in Integer Linear Programming (ILP). Though an ILP problem can be solved with an ILP solver, the problem is NP-hard and it is difficult to obtain the exact solution in situations where immediate responses are needed. Our solution is to propose optimization heuristics that exploit Lagrangian relaxation to obtain good approximate solutions within feasible computation times. Experiments on the document understanding conference 2004 (DUC'04) dataset show that our Lagrangian relaxation based heuristics completes in feasible computation time but achieves higher ROUGE scores than state-of-the-art approximate methods.
KOSERUBE (version 1) is a system that automatically generates stories and discourses in the style of a folk tale with its characters, places and objects relating to Iwate Prefecture. The system also expresses sentences and music, and automatically edits visual objects relating to the generated narratives. The user can operate and appreciate the process through a visual interface. In the context of our narrative generation system research, KOSERUBE mainly uses Propp-based story grammar as the major automatic generation mechanism. In this paper, first, we give a system overview covering the user interface and the narrative generation mechanism depending on Propp-based story grammar. Next, we present the results of a questionnaire survey and a demonstration at an exhibition to identify relevant issues and discuss the expansion of the system to the second version. One of the important issues identified is the processing of the narratives for consistency and naturalness. In the questionnaire survey results, both positive and negative opinions about jumps or gaps in a narrative appeared. For the development of version 2, our aim is to implement both mechanisms for managing narrative consistency and those for deviating from it. We also discuss other topics regarding the next version. Overall, we propose a new type of multimedia content system with an automatic narrative generation system.