Dongxu Zhang 2015-01-19

来自cslt Wiki
2015年1月19日 (一) 13:37Zhangdx讨论 | 贡献的版本

(差异) ←上一版本 | 最后版本 (差异) | 下一版本→ (差异)
跳转至: 导航搜索

Accomplished this week

  • use yago for training
  • get comparable performance by tree training and link training. And get a higher performance by joint training them.

Planned for next week

  • use text information and train word vector together.
  • employ a more reasonable method to train the tree.
  • try to train on the whole data set.