Evaluation of unsupervised semantic mapping of natural language with Leximancer concept mapping

Abstract

The Leximancer system is a relatively new method for transforming lexical co-occurrence information from natural language into semantic patterns in an unsupervised manner. It employs two stages of co-occurrence information extraction-semantic and relational-using a different algorithm for each stage. The algorithms used are statistical, but they employ nonlinear dynamics and machine learning. This article is an attempt to validate the output of Leximancer, using a set of evaluation criteria taken from content analysis that are appropriate for knowledge discovery tasks.

The past is prologue: Moving on from Culture’s Consequences

“…As noted by Campbell et al (2011, p. 92), as "evidence accumulates … The tool automatically and efficiently learns that words predict which concepts." Leximancer derived concept identification has exhibited high face validity, high reliability and reproducibility and thematic clustering without facing the biases associated with manually coded text analyses (Smith & Humphreys, 2006). It is found to be superior to other approaches, such as NVIVO, particularly when the goal is to quickly discover key themes and the relationship between concepts (Sotiriadou, Brouwers & Le, 2014).…”

25+ Years of Business Intelligence and Analytics Minitrack at HICSS: A Text Mining Analysis

“…Due to its versatile applications by other researchers, Leximancer has been described as a "textmining tool for visualizing the structure of concepts and themes in text" [2:25], a visual tool for making sense of big data [3], a data mining tool [36], a tool for qualitative data analysis [11], quantitative content analysis [28] and a quantitative tool for conducting qualitative analysis of text data [21].…”

“…The resulting concept maps are used by Leximancer to establish the relational strength between different concepts so that it could be used by the researchers to interpret the strength of different associations. The software has been extensively evaluated for stability, reproducibility and correlative validity of underlying statistical algorithms [36].…”

Educational Technology trends in the Ibero-American world: 20 years of the Edutec-e journal

“…The result of this qualitative analysis with Leximancer is a set of concept maps, where the concept frequency, the hierarchical order of appearance and the proximity among concepts are visualised (Smith & Humphreys, 2006). Each thematic region is formed based on the connectedness of concepts and highlighted by the most relevant concept in terms of frequency and connections (relational analysis).…”