科研之友
微信
新浪微博
Facebook
分享链接
摘要
Neurodegenerative disorders such as Parkinson's and Alzheimer's disease are devastating and costly illnesses, a source of major global burden. In order to provide successful interventions for patients and reduce costs, both causes and pathological processes need to be understood. The ApiNATOMY project aims to contribute to our understanding of neurodegenerative disorders by manually curating and abstracting data from the vast body of literature amassed on these illnesses. As curation is labour-intensive, we aimed to speed up the process by automatically highlighting those parts of the PDF document of primary importance to the curator. Using techniques similar to those of summarisation, we developed an algorithm that relies on linguistic, semantic and spatial features. Employing this algorithm on a test set manually corrected for tool imprecision, we achieved a macro F-1-measure of 0.51, which is an increase of 132% compared to the best bag-of-words baseline model. A user based evaluation was also conducted to assess the usefulness of the methodology on 40 unseen publications, which reveals that in 85% of cases all highlighted sentences are relevant to the curation task and in about 65% of the cases, the highlights are sufficient to support the knowledge curation task without needing to consult the full text. In conclusion, we believe that these are promising results for a step in automating the recognition of curation-relevant sentences. Refining our approach to pre-digest papers will lead to faster processing and cost reduction in the curation process.
- 出版日期2017-3-27
- 单位南京信息工程大学
