• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文会议>Text processing >Identification of Relevance and Support for Consumer Health Information
【24h】

Identification of Relevance and Support for Consumer Health Information

机译：识别消费者健康信息的相关性和支持

获取原文
获取原文并翻译 | 示例
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

With a rapid growth of queries posted for the search on internet that were related to medical information raised the need of acquiring the right and relevant information related to those queries. The information systems available at present are providing the documents that matches the user query, but to check whether they really matches the queries is becoming a difficult task for the layman. This is because the consumer does not have any knowledge related to Medical records and its nomenclature. Consumer Health Information Search (CHIS) is a track organized to cater the need of medical information search. As a part of this track two tasks were designed. (1) Given a query and a document containing a set of sentences, the task is to identify whether the sentence selected is relevant/irrelevant to the query posted. (2) To identify whether the sentence selected from the document is supporting the query or opposing the query or is in neutral state. The solution of Task_1 is achieved by selecting the similarity scores as a feature. The mean of this similarity scores were computed to identify the relevant nature of the sentence to the query. Task 2 is viewed as a multiple classification problem and is solved by making use of C-Support Vector Machine Classifier. The model was tested on data set provided by the CHIS track organizers. The results obtained by our model that was designed for task_1 were not to the satisfactory level when compared to the results of other track participants. For Task 2 C-support vector machine model is applied. In that Tf-idf score is used as a feature for the model. Results obtained for task_2 obtained the highest accuracy scores when compared with the other models submitted by different participants of the track.
机译：随着在互联网上发布的与医学信息有关的查询的快速增长，提出了获取与这些查询相关的正确和相关信息的需求。当前可用的信息系统正在提供与用户查询匹配的文档，但是要检查它们是否确实与查询匹配，已成为外行的一项艰巨任务。这是因为消费者不了解任何有关病历及其术语的知识。消费者健康信息搜索（CHIS）是为满足医疗信息搜索需求而组织的。作为该跟踪的一部分，设计了两个任务。 （1）给定一个查询和一个包含一组句子的文档，任务是识别所选句子是否与发布的查询相关/无关。 （2）识别从文档中选择的句子是支持查询还是反对查询或处于中立状态。通过选择相似性分数作为特征来实现Task_1的解决方案。计算该相似性分数的平均值，以识别该句子与查询的相关性质。任务2被视为多重分类问题，可以通过使用C支持向量机分类器解决。该模型在CHIS赛道组织者提供的数据集上进行了测试。与其他跟踪参与者的结果相比，我们为任务_1设计的模型获得的结果未达到令人满意的水平。对于任务2，使用C支持向量机模型。在这种情况下，Tf-idf分数用作模型的特征。与轨道不同参与者提交的其他模型相比，为task_2获得的结果获得了最高的准确性得分。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号