552 Abstract—In many documents such as admission form, bank cheques, memorandums, letters and application forms machine printed and handwritten characters are mixed. Since the algorithms for recognition of machine-printed texts and handwritten texts are different, it is necessary to distinguish between these two types of texts before giving it to respective OCR systems to process it. This separ...

Current Chinese social media text summarization models are based on an encoderdecoder framework. Although its generated summaries are similar to source texts literally, they have low semantic relevance. In this work, our goal is to improve semantic relevance between source texts and summaries for Chinese social media summarization. We introduce a Semantic Relevance Based neural model to encoura...

This paper presents ongoing work dedicated to parsing the textual structure of procedural texts. This work is based on a two level approach: first, basic structures in texts are segmented using automata techniques and then a text grammar is applied to assemble these fragments. The grammar is based on X-bar syntax, transposed to text structure.

Text tagging is a very important tool for various applications in natural language processing, namely the morphological and syntactic analysis of texts, indexation and information retrieval, "vocalization" of Arabic texts, and probabilistic language model (n-class model). However, these systems based on the lexemes of limited size, are unable to treat unknown words consequently. To overcome thi...

E-learning is popularized so fast and Collaborative Learning (CL) becomes so important an instructional strategy. There are huge Group Session (GS) texts needed to be analyzed to evaluate CL, thus the automatic or semi-automatic methods of analyzing the GS texts become very important. In this paper we present a method called Interaction Analysis depended on Knowledge Extraction (IAKE) to analyz...

This paper presents some improvements that have been made to UNLSumm, an automatic summarizer of UNL Texts. These are texts encoded in the Universal Networking Language (UNL), an artificial language intended to convey conceptualizations of natural language texts. In UNLSumm, a set of heuristics is responsible for, based in the UNL characteristics, prune sentences of a UNL Text, generating its U...

The development of the XML and RDF(S) standards offer a positive environment for machine learning to enable the automatic XML-annotation of texts that can encourage the extension of Semantic Web applications. After reviewing the current limitations of information extraction technology, specifically its lack of portability to new domains, we introduce the PIA project for automatically XML-annota...

In this article is presented a tool for labeling bilingual parallel corpora and aligning texts using heuristics based on word frequency, position and lexicographical similarity (using Edit Distance). Bilingual corpora annotation is a very laborious task but essential at the time of developing knowledge bases for the accomplishment of automatic translations between different languages. This tool...