Fig 1 - available via license: Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International
Content may be subject to copyright.
An example of a news article with well-formed text and an example of an academic resume with free-form text. All the person names are highlighted.
Source publication
Person names are essential entities in the Named Entity Recognition (NER) task. Traditional NER models have good performance in recognising well-formed person names from text with consistent and complete syntax, such as news articles. However, user-generated documents such as academic homepages and articles in online forums may contain lots of free...
Contexts in source publication
Context 1
... NER models [6], [7], [8] have achieve good performance on recognising person names from well-formed text with consistent and complete syntax, such as news articles (cf. Figure 1(a) ). Person names in such text often have straightforward patterns, e.g., first name in full followed by last name in full. ...
Context 2
... may appear in many applications, such as user-generated academic homepages, resumes, articles in online forums and social media (cf. Figure 1(b)). These text often contain person names of various forms with incomplete syntax. ...
Context 3
... first study the impact of using our fine-grained annotations with different fusion strategies ( Figure 10) and different models. Specifically, four fusion strategies are tested: ...
Context 4
... can see from the Figure 13, 14 that our proposed model can not only recognize more person names than Stanford NER, our model also provide fine-grained name forms with higher accuracy. We also test two models on short text, Figure 11, 12 show that our model also perfroms well on text without too much context, and it can distinguish the same tokens with different meanings. ...
Context 5
... in organization names: Figure 15 shows a typical error that the model labels the name of an organization as a person name. Even though the organization is named by a person "Ronald O. Perelman", we expecte that the model could distinguish organization names from person names by the context information, such as the "Department of Emergency Medicine". ...
Context 6
... in well-formatted text: Figure 16 shows a typical error that the model does not recognize some person names in the well-formatted text, such as in the introductions or in the career descriptions. The reason is that, in the FinegrainedName dataset, most person names appear in the publications and come with different formats as those the introductions. ...
