A Framework for Contextual Named Entity Linking

上下文命名实体链接的框架

• 批准号: 521169-2017
• 负责人: Chiang, FeiYen
• 金额: $ 1.82万
• 依托单位: McMaster University
• 依托单位国家: 加拿大
• 项目类别: Engage Grants Program
• 财政年份: 2017
• 资助国家: 加拿大
• 起止时间: 2017-01-01 至 2018-12-31
• 项目状态: 已结题
• 来源: https://www.nserc-crsng.gc.ca/ase-oro/Details-Detailles_eng.asp?id=640115
• 关键词: Framework; Contextual; Named; Entity; Linking

This project aims to develop a context aware entity linking framework for VerticalScope that improves theaccuracy of extracted entities, and links them to true entities in a database. We will investigate and developsolutions to extract mentions of entities from forum posts by considering the overall context and semantics oftext, linking extracted entities to a product taxonomy or database, and evaluate the accuracy of the overallmodel. Given the high cost to purchase vertical-specific catalogues, we explore clustering techniques toidentify clusters of words that represent a named entity and link these mentions to enrich a product taxonomyfor a class of vertical domains. The developed technology will enable VerticalScope to glean deeper insightson current product trends in forum post discussions, drive more timely decision making, and increase the valueof their forum assets to business vendors.

该项目旨在为VerticalScope开发一个上下文感知实体链接框架，以提高提取实体的准确性，并将它们链接到数据库中的真实实体。我们将研究和开发解决方案，通过考虑文本的整体上下文和语义，将提取的实体链接到产品分类法或数据库，并评估整体模型的准确性，从论坛帖子中提取实体的提及。考虑到购买垂直特定目录的高成本，我们探索聚类技术来识别代表命名实体的单词簇，并将这些提及链接起来，以丰富垂直领域类的产品分类。开发的技术将使VerticalScope能够在论坛帖子讨论中更深入地了解当前产品趋势，推动更及时的决策，并增加他们的论坛资产对商业供应商的价值。