In this paper, we consider the extraction of speaker identity from audio records of broadcast news without a priori acoustic information about speakers. Using an automatic speech recognition system and an automatic speaker diarization system, we present improvements for a method which allows to extract speaker identities from automatic transcripts and to assign them to speech segments.Experiments are carried out on French broadcast news records from the ESTER 1 evaluation campaign. Experimental results using outputs of automatic speech recognition and automatic diarization are presented.

No abstract

International audienceRecently there has been an increase in interest towards clustering short text because it could be used in many NLP applications. According to the application, a variety of short text could be defined mainly in terms of their length (e.g. sentence, paragraphs) and type (e.g. scientific papers, newspapers). Finding a clustering method that is able to cluster short text in general is difficult. In this paper, we cluster 4 different corpora with different types of text with varying length and evaluate them against the gold standard. Based on these clustering experiments, we show how different similarity measures, clustering algorithms, and cluster evaluation methods effect the resulting clusters. We discuss four existing corpus based similarity methods, Cosine similarity, Latent Semantic Analysis, Short text Vector Space Model, and Kullback-Leibler distance, four well known clustering methods, Complete Link, Single Link, Average Link hierarchical clustering and Spectral clustering, and three evaluation methods, clustering F-measure, adjusted Rand Index, and V. Our experiments show that corpus based similarity measures do not significantly affect the clusters and that the performance of spectral clustering is better than hierarchical clustering. We also show that the values given by the evaluation methods do not always represent the usability of the clusters

Named Entity Recognition (NER) from speech usually involves two sequential steps: transcribing the speech using Automatic Speech Recognition (ASR) and annotating the outputs of the ASR process using NER techniques. Recognizing named entities in automatic transcripts is difficult due to the presence of transcription errors and the absence of some important NER clues, such as capitalization and punctuation. In this paper, we describe a methodology for speech NER which consists of incorporating NER into the ASR process so that the ASR system generates transcripts annotated with named entities. The combination is achieved by adapting ASR language models and pre-annotating the pronunciation dictionary. We evaluate this method on ESTER 2 corpus, and show significant improvements over traditional approaches.

Bonom is a distributed agent based system to perform information retrieval on the Internet. The middle agents which perform the matching between site agents (information providers) and user agents (requesters) are numerous, and can be world wide distributed. Moreover; the agents are structured into nested communities according to a hierarchy of information domains. This paper focuses on a web site indexing process which uses disambiguated ontologies. It allows the site agent to know to which communities it has to subscribe to, in order to get requests it can accurately process.

We address in this paper the issue of text reuse in liturgical manuscripts of the middle ages. More specifically, we study variant readings of the Obsecro Te prayer, part of the devotional Books of Hours often used by Christians as guidance for their daily prayers. We aim at automatically extracting and categorising pairs of words and expressions that exhibit variant relations. For this purpose, we introduce a linguistic classification that allows to better characterize the variants than edit operations. Then, we study the evolution of Obsecro Te texts from a temporal and geographical axis. Finally, we contrast several unsupervised state-of-the-art approaches for the automatic extraction of Obsecro Te variants. Based on the manual observation of 772 Obsecro Te copies which show more than 21,000 variants, we show that the proposed methodology is helpful for an automatic study of variants and may serve as basis to analyse and to depict useful information from devotional texts.

No abstract

No abstract