情景驱动的机会发现关键技术研究

项目名称： 情景驱动的机会发现关键技术研究

项目编号： No.61303164

项目类型： 青年科学基金项目

立项/批准年度： 2014

项目学科： 自动化技术、计算机技术

项目作者： 王浩

作者单位： 中国科学院软件研究所

项目金额： 26万元

中文摘要： 现实世界中，对罕见重要事件或者形势的发现往往对人的判断决策起着更重要的作用。机会发现作为扩展知识发现和数据挖掘的一个新兴研究领域应运而生。机会发现是通过人机交互过程来发现对未来决策有重要影响的潜在事件或者形势。然而，在实际应用中仍然存在如下三个问题：①序列事件的时间性被忽略导致机会挖掘精度较低；②人机交互过程中人对机会感知程度有限；③日趋复杂的动态环境迫使机会发现过程的效率和自适应性有待于进一步提高。本项目拟以"情景"为驱动，研究基于序列事件的机会挖掘方法以及情景图构建方法；研究面向情景感知的机会检索模型；研究基于动态情景的人机协同机会发现方法；最后开发原型系统并实例验证分析。研究成果将扩展传统的“模式-预测”发现模式，为“情景-预见”发现过程提供关键技术支撑，可广泛应用于自然灾害早期预警、潜伏病情早期诊断、客户隐性需求开发和潜在商机发现等。本项目也促进了这个崭新的学科在我国的快速发展。

中文关键词： 机会发现；情景图；人机协同；机会度量；罕见事件

英文摘要： In the real world, discovering rare but important events or situations is playing a more vital role in human decision making. Chance Discovery has been developed as an emerging field, extending Knowledge Discovery and Data Mining, which is a human-computer interaction process that detects rare but important chances for decision making. However, the following three problems still need to be solved in the practical applications: (1) ignoring the timeliness of sequence of events causes lower precision of chance mining; (2) human makes limit level of awareness during the human-computer interaction; (3) The efficiency and adaptability of chance discovery process needs to be further improved in an increasingly complex and dynamic environment. This project takes "scenario" as the driving power, researching on the approach of sequence-based chance mining and scenario construction, scenario awareness oriented chance retrieval model, and human-computer collaborative chance discovery approach under the dynamic scenario. At last, a prototype system will be developed and validated by case studies. The result will extend traditional "Pattern-Prediction" discovery and provide the key techniques for "Scenario-Foresight" discovery process, which can be widely used in the early warning of natural catastrophes, early diagnosis of

英文关键词： Chance Discovery；Scenario Graph；Human-machine Cooperation；Chance Metric；Rare Event