| Presentation | 2017-02-10 Extractive Summarization of Financial Statement Using Multi-Task Learning Masaru Isonuma, Toru Fujino, Jumpei Ukita, Haruka Murakami, Kimitaka Asatani, Junichiro Mori, Ichiro Sakata, |
|---|---|
| PDF Download Page |  PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | In this paper, we proposed a methodology of summarizing financial statements which contributes to high quality investment decision-making. In the task of supervised extractive summarization of financial statements, the lack of training data is crucial issue. To solve the issue, we propose a extractive summarization architecture using multi-task learning with financial results prediction. The sentences focused in financial results prediction correspond to the sentences that should be extracted, therefore the learning on financial results prediction contributes to the sentence extraction task. The experiment shows that our model improves the accuracy of summarization especially in the case of few training data. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | Sentence Extraction / Financial Report / Financial Results Prediction / Multi-Task Learning / Recurrent Neural Networks |
| Paper # | NLC2016-47 |
| Date of Issue | 2017-02-02 (NLC) |
| Conference Information | |
| Committee | NLC / IPSJ-IFAT |
|---|---|
| Conference Date | 2017/2/9(2days) |
| Place (in Japanese) | (See Japanese page) |
| Place (in English) | |
| Topics (in Japanese) | (See Japanese page) |
| Topics (in English) | |
| Chair | Hiroshi Kanayama(IBM) |
| Vice Chair | Makoto Ichise(NTT DoCoMo) / Takeshi Sakaki(Univ. of Tokyo/Hottolink) |
| Secretary | Makoto Ichise(Ryukoku Univ.) / Takeshi Sakaki(Kyushu Inst. of Tech.) |
| Assistant | Ryuichiro Higashinaka(NTT) / Mitsuo Yoshida(Toyohashi Univ. of Tech.) |
| Paper Information | |
| Registration To | Technical Committee on Natural Language Understanding and Models of Communication / Special Interest Group on Information Fundamentals and Access Technologies |
|---|---|
| Language | JPN |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | Extractive Summarization of Financial Statement Using Multi-Task Learning |
| Sub Title (in English) | |
| Keyword(1) | Sentence Extraction |
| Keyword(2) | Financial Report |
| Keyword(3) | Financial Results Prediction |
| Keyword(4) | Multi-Task Learning |
| Keyword(5) | Recurrent Neural Networks |
| 1st Author's Name | Masaru Isonuma |
| 1st Author's Affiliation | University of Tokyo(UTokyo) |
| 2nd Author's Name | Toru Fujino |
| 2nd Author's Affiliation | University of Tokyo(UTokyo) |
| 3rd Author's Name | Jumpei Ukita |
| 3rd Author's Affiliation | University of Tokyo(UTokyo) |
| 4th Author's Name | Haruka Murakami |
| 4th Author's Affiliation | University of Tokyo(UTokyo) |
| 5th Author's Name | Kimitaka Asatani |
| 5th Author's Affiliation | University of Tokyo(UTokyo) |
| 6th Author's Name | Junichiro Mori |
| 6th Author's Affiliation | University of Tokyo(UTokyo) |
| 7th Author's Name | Ichiro Sakata |
| 7th Author's Affiliation | University of Tokyo(UTokyo) |
| Date | 2017-02-10 |
| Paper # | NLC2016-47 |
| Volume (vol) | vol.116 |
| Number (no) | NLC-451 |
| Page | pp.pp.45-50(NLC), |
| #Pages | 6 |
| Date of Issue | 2017-02-02 (NLC) |