Paper Abstract and Keywords |
Presentation |
2012-08-30 13:40
Extraction of Sentences Describing Problems from Sales Force Management System Texts. Daigo Sugihara, Tomoko Ohkuma, Koji Satake, Yasuhide Miura, Keigo Hattori, Hiroshi Masuichi (Fuji Xerox) NLC2012-11 |
Abstract |
(in Japanese) |
(See Japanese page) |
(in English) |
Sales persons are recommended to know customers' problems, because knowing customers' problems will lead to better proposals for customers. Knowing problems about services or products provided to their customers will lead to improvement of their services or products. Therefore we tried to extract sentences describing problems from texts in sales force management system, with aim of analyzing various problems or searching problems and solution in the past. Surveying texts in sales force management system, we found out that definition for sentences describing problems we should extract is "Sentences described as object or result for solution, such as troubled situation or desired goal". We defined extraction of sentences describing problems as classification whether input sentence is describing problems or not. Our experimental result showed that our SVM classifier achieved F-measure of 0.40 and the most effective features were those about vocaburaries such as negative expressions from polarity dictionaries and function words representing demand or capability. |
Keyword |
(in Japanese) |
(See Japanese page) |
(in English) |
SVM / Binary Classificatin / Sentence Extraction / Sales Force Management System / Customers' Problems / / / |
Reference Info. |
IEICE Tech. Rep., vol. 112, no. 196, NLC2012-11, pp. 7-12, Aug. 2012. |
Paper # |
NLC2012-11 |
Date of Issue |
2012-08-23 (NLC) |
ISSN |
Print edition: ISSN 0913-5685 Online edition: ISSN 2432-6380 |
Copyright and reproduction |
All rights are reserved and no part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording, or any information storage and retrieval system, without permission in writing from the publisher. Notwithstanding, instructors are permitted to photocopy isolated articles for noncommercial classroom use without fee. (License No.: 10GA0019/12GB0052/13GB0056/17GB0034/18GB0034) |
Download PDF |
NLC2012-11 |
|