You are here

evaluating

The use of an indexing language in the catalogues of university libraries: a method for its evaluation by users based on a sociocognitive approach
Presentation Type: 
Poster
Language: 
English
Presentation Paper: 
Music indexing and retrieval: evaluating the social production of music metadata and its use

This article will focus on music indexing and retrieval from different points of view. Four elements will be examined: music metadata, indexing and retrieval methods (classic indexing, collaborative indexing and social tagging), tools and users. Regarding the users, we will look at their access modes, their possible participation in indexing, and their music information-seeking behaviors. Do they look for music information in pay-for-music sites (iTunes), radio stations, blogs, or social networks (MySpace, YouTube, etc)? The question we raise is which music information systems would be suitable in a social web era. An evaluation of the existing indexing and retrieval modes was conducted. It was based on both quantitative and qualitative approaches. Our research methodology used interviews, online questionnaires and semi-directed questionnaires. We believe that the results of our evaluation will be useful for music information indexing in a Socio-Semantic Web context.

Presentation Type: 
Talk
Language: 
English
Presentation Audio: 
Presentation Visual: 
Presentation Paper: 
Semantic metadata annotation: tagging Medline abstracts for enhanced information access

Purpose - The object of this study is to develop methods for automatically annotating the argumentative role of sentences in scientific abstracts. Working from Medline abstracts, we classified sentences into four major argumentative roles: objective, method, result, conclusion. The idea is that if the role of each sentence can be marked up, then this metadata can be used during information retrieval to seek for particular types of information such as novelty, conclusions, methodologies, aims/goals of a scientific piece of work. Methodology - Two approaches were tested: linguistic cues and positional heuristics. Linguistic cues are lexico-syntactic patterns modeled as regular expressions implemented in a linguistic parser. Positional heuristics make use of the relative position of a sentence in the abstract to deduce its argumentative class. Findings - Our experiments showed that positional heuristics attained a much higher degree of accuracy on Medline abstracts with an F-score of 64% whereas the linguistic cues only attained an F-score of 12%. This is mostly because sentences from different argumentative roles are not always announced by surface linguistic cues. Research limitations/implications - A limitation to this study is that we were not able to test other methods to perform this task such as machine learning techniques which have been reported to perform better on Medline abstracts. Also, to compare the results of our study to earlier studies using Medline abstracts, the different argumentative roles present in Medline had to be mapped onto four major argumentative roles. This may have favorably biased the performance of the sentence classification by positional heuristics. Originality/value. To the best of our knowledge, our study presents the first instance of evaluating linguistic cues and positional heuristics on the same corpus

Presentation Type: 
Talk
Language: 
English
Presentation Audio: 
Presentation Visual: 
Presentation Paper: 
Subscribe to evaluating