Semantic Indexing and Typed Hyperlinking
نویسنده
چکیده
In this paper, we describe linguistically sophisticated tools for the automatic annotation and navigation of on-line documents. Creation of these tools relies on research into finite-state technologies for the design and development of lexicallyintensive semantic indexing, shallow semantic understanding, and content abstraction techniques for texts. These tools utilize robust language processing techniques to generate multi-purpose data structures called LEXICAL WEBS, used in the system TEXTRACT, an automated semantic indexing program designed to parse, index, and hyperlink electronic documents.
منابع مشابه
Semantic Indexing with Typed Terms Using Rapid Annotation
Illustrated with an example of a manual semantic resource for German, the use of typed index terms for semantic indexing is proposed. Types group index terms in the same semantic category and can be used by any search or cluster mechanism. To obtain semantic resources, a method to rapidly annotate large corpora is described in detail.
متن کاملHNews: An Enhanced Multilingual Hyperlinking News Platform
In this paper, we describe the HNews platform, a Web-based system addressing the general problem of aggregating and enriching news from different sources and languages. In the indexing stage, the news items gathered from RSS feeds or video streams are analyzed through Information Extraction tools. Their topical category information and the Named Entities mentions are recognized and used to crea...
متن کاملFXPAL Experiments for TRECVID 2016 Video Hyperlinking
This is a summary of our participation in the TRECVID 2016 video hyperlinking task (LNK). We submitted four runs in total. A baseline system combined on established vectorspace text indexing and cosine similarity. Our other runs explored the use of distributed word representations in combination with fine-grained inter-segment text similarity measures.
متن کاملGhent University-IBBT at MediaEval 2012 Search and Hyperlinking: Semantic Similarity using Named Entities
In this paper, we attempt to tackle the MediaEval 2012 Search and Hyperlinking challenge, which focuses on video segment retrieval from a large dataset, based on short natural language queries, as well as linking the resulting segments to related ones. Our approach makes use of three semantic similarity metrics, merged by applying late fusion.
متن کامل