“141003-progress”版本间的差异
来自cslt Wiki
（→resourc） |
|||
| （相同用户的2个中间修订版本未显示） | |||
| 第11行： | 第11行： | ||
* may using simple parser like POS | * may using simple parser like POS | ||
* Add more nlp method to system | * Add more nlp method to system | ||
| + | ==patents== | ||
| + | * GA method that added in QA system | ||
| + | * How to protect the patents for company, may ask help with agent | ||
| + | ==resource== | ||
| + | slider[http://cslt.riit.tsinghua.edu.cn/mediawiki/images/b/bb/HuilanQA.pdf] | ||
2014年10月15日 (三) 04:48的最后版本
will discuss after national day. later will update the detail ppt
Structure Knowledge
- labeling system designed to fit the function
- define the structure for government affairs.
- design the new framework to add the function, which combine the fuzzy-mach with the knowledge base(like Triples for bookqa).
- talk more about the next framework's version and labeling system.
Semantic Similarity
- may need to train the word2vec for specific area.
- test the how-net and synonymy substitution of chinese text.
Template Expansion
- may using simple parser like POS
- Add more nlp method to system
patents
- GA method that added in QA system
- How to protect the patents for company, may ask help with agent
resource
slider[1]
