ReaderBench : An Integrated Cohesion-Centered Framework

Abstract

Abstract.ReaderBench is an automated software framework designed to support both students and tutors by making use of text mining techniques, advanced natural language processing, and social network analysis tools. ReaderBench is centered on comprehension prediction and assessment based on a cohesion-based representation of the discourse applied on different sources (e.g., textual materials, behavior tracks, metacognitive explanations, Computer Supported Collaborative Learning -CSCL -conversations). Therefore, ReaderBench can act as a Personal Learning Environment (PLE) which incorporates both individual and collaborative assessments. Besides the a priori evaluation of textual materials' complexity presented to learners, our system supports the identification of reading strategies evident within the learners' self-explanations or summaries. Moreover, ReaderBench integrates a dedicated cohesion-based module to assess participation and collaboration in CSCL conversations.

Extracting Gamers' Opinions from Reviews

“…Like Crossley et al, we opted to integrate multiple linguistic resources consisting of various word lists or vectors used in general text classification, as well as words with particular semantic valence in accordance to predefined taxonomies. Even from the beginning we must highlight the main differentiator of this approach which is designed to provide multi-lingual sentiment analyses integrated in the ReaderBench framework (http://readerbench.com/, [11][12][13]). As stated in the introduction, this paper presents a pilot study for extracting gamers' opinions expressed in English language that will be later on extended using the ReaderBench framework which already supports multiple languages including English, French and Romanian, as well as partial support for Spanish, Dutch, Italian and Latin.…”

What Makes Your Writing Style Unique? Significant Differences Between Two Famous Romanian Orators

“…In order to explore the oratorical styles of both Romanian personalities, we rely on a previously validated textual complexity model, integrated in our ReaderBench framework [11][12][13][14], adapted for Romanian language [12], that addresses multiple facets of text difficulty and comprehension [11]: text features (e.g., length, structure or use of punctuation) [15], textual formality (e.g., vocabulary, slang, phrasal verbs, use of idiomatic language, and so on) [16], and textual styles (e.g., simple/complex sentences, stylistic markers, cohesion, etc.) [17].…”

Building a semantic recommendation engine for news feeds based on emerging topics from tweets

“…Unfortunately, the task at hand increases in difficulty for loosely-coupled structures such as word sequences (i.e., text segments with few tokens created and employed by human individuals). In this paper, we use various Natural Language Processing (NLP) techniques, including semantic distances in lexicalized ontologies, Latent Semantic Analysis (LSA) vector models and Latent Dirichlet Allocation (LDA) topic distributions, all integrated within our ReaderBench framework [4][5][6][7] in order to extract the key concepts from a Tweet corpus. Furthermore, we attempt to match valid news feeds extracted via crawling to two major news broadcasting systems: CNN and BBC.…”