“Hulan-2015-03-05”版本间的差异

来自cslt Wiki
跳转至: 导航搜索
Online Learning
Lr讨论 | 贡献
Frame Work
 
(相同用户的9个中间修订版本未显示)
第11行: 第11行:
 
* demo (done)
 
* demo (done)
 
===Online Learning===
 
===Online Learning===
* v1 done[[文件:Tianyi's_online_learning--20150305.pdf]]
+
* v1 done [http://cslt.riit.tsinghua.edu.cn/mediawiki/images/1/14/Tianyi%27s_online_learning--20150305.pdf]
 
* v2
 
* v2
:* add the query relevancy from user click the recommended list
+
:* add the query relevancy from user click the recommended list(hold)
  
 
===reverse module===
 
===reverse module===
* knowledge graph
+
* v1 demo
:* process of setting knowledge graph and support reverse ask[http://cslt.riit.tsinghua.edu.cn/mediawiki/images/1/18/%E6%94%BF%E5%BA%9C%E7%BB%84%E7%BB%87%E6%9C%BA%E6%9E%84%E5%9B%BE%E8%B0%B1--%E6%B1%87%E8%81%94.pdf]
+
:* simple demo is done, solve the Chinese problem.
:* search the opensource jena and D2R
+
 
:* give a demo
+
==Frame Work==
::* simple demo is done
+
* develop the code in different module
 +
:* process  
 +
::* search module : lucene search , knowledge graph search => lr
 +
::* Session management module => lr
 +
::* offline learn module => tianyi
 +
::* online learn module =>tianyi
 +
:* NLP tool
 +
::* Entity recognition and NER =>lr
 +
::* Spell check module
 +
::* word similarity module
 +
:* common module:  pass parameters in different module
 +
* problem to solve
 +
:* define the common inference
  
 
==Knowledge Management and labeling system==
 
==Knowledge Management and labeling system==
* continue coding.
+
* caoli ready to label the liangshan-online data
 +
 
 
==Patent==
 
==Patent==
* done
 
==new word find==
 
 
* done
 
* done
 
==leftover problem==
 
==leftover problem==

2015年3月5日 (四) 04:20的最后版本

Dialog system

Algorithm

Spell mistake

  • done

improve fuzzy match

  • hold

improve lucene search

  • learning to rank [1]
  • result: 0.688->0.78

Multi-Scene Recognition

  • demo (done)

Online Learning

  • v1 done [2]
  • v2
  • add the query relevancy from user click the recommended list(hold)

reverse module

  • v1 demo
  • simple demo is done, solve the Chinese problem.

Frame Work

  • develop the code in different module
  • process
  • search module : lucene search , knowledge graph search => lr
  • Session management module => lr
  • offline learn module => tianyi
  • online learn module =>tianyi
  • NLP tool
  • Entity recognition and NER =>lr
  • Spell check module
  • word similarity module
  • common module: pass parameters in different module
  • problem to solve
  • define the common inference

Knowledge Management and labeling system

  • caoli ready to label the liangshan-online data

Patent

  • done

leftover problem