- 分享主题:Feature Fusion
- 论文标题:Learning Effective Representations for Person-Job Fit by Feature Fusion
- 论文链接:https://arxiv.org/pdf/2006.07017.pdf
- 分享人:唐共勇

1. Summary

【必写】,推荐使用 grammarly 检查语法问题,尽量参考论文 introduction 的写作方式。需要写出

  1. 这篇文章解决了什么问题?
  2. 作者使用了什么方法(不用太细节)来解决了这个问题?
  3. 你觉得你需要继续去研究哪些概念才会加深你对这篇文章的理解?

This article attempts to solve the “person job” problem, which refers to the use of machine learning algorithm to match candidates and positions on the online recruitment platform. The performance of the algorithm largely depends on the learning representation of candidates and positions. This paper proposes to learn the comprehensive and effective representation of candidates and jobs through feature fusion. First, extract the resume and job requirements to obtain certain information, and then learn the historical information of candidates and publishers to try to learn the implied representation of employment or acceptance intention. The two parts of features are fused to represent the “person job” problem. The experimental results show that both explicit and implicit features have a positive effect on the final result.This article makes price predictions based on high-frequency trading data, which is different from the traditional prediction of rising and falling. The author completes this task as a regression task.

2. 你对于论文的思考

需要写出你自己对于论文的思考,例如优缺点,你的takeaways

优点:

  1. job推荐类文章,提出了一种处理简历与岗位之间特征表示的方法
  2. 准确性和可解释性高,能够捕捉岗位申请者与发布者的隐含意图
  3. 文章架构很好,整个逻辑十分清晰

缺点:

  1. 应用面小,应用场景固定

    3. 其他

    【可选】

基本问题定义
image.png
显示特征学习
image.png
隐式特征学习
image.png