Python-远程-1个月

项目编号：【143985】
主要需求：
1. 爬虫：抓取us patent center特定xml文件
2. 提取文本：从xml文件中提取文字信息→doc
3. NER & NED/NEL：从doc中提取knowledge concepts（entities in some classes in Wikidata），可采用以下几种做法
1. annotate a training set + finetune a NED model (e.g. BLINK, LUKE on github)
2. annotate a training set + PatentBert + train NER using DYGIE++(on github)
3. spacy pipeline: annotate a training set + PatentBert train NER + Wiki NEL
4. spacy pipeline: annotate a training set + PatentBert train NER + entity clustering
5. joint NER & NED using merged knowledge base

非必要需求：
1. Build a two-mode network of the extracted entities and put them into Neo4J

语言为python，model基本都有GitHub repo和论文进行参考，可调用google cloud GPU，只提供code就好，training可以我方来做