The solution dedicated to textual data processing
The Sphinx Quali option is the solution dedicated to the processing of textual data and semantic analysis. It integrates recent advances in knowledge engineering (ontology, semantic networks, etc.) and is at the forefront of innovative technologies.
With the expansion of data of all kinds and the multiplication of opinions on the web and social networks, it becomes essential to process data from whatever source: open-ended questions, interviews, focus group, press articles, extracts from websites, comments from forums and social networks ... Sphinx Quali makes it possible to import all of these corpuses.
Examples of qualitative studies and methods
Satisfaction survey - Hotline
This customer satisfaction survey conducted among the customers of a well-known software publisher complements the service quality measurements with questions on the problems encountered and the justification of the evaluations made. Keywords and recurring expressions confirm the good level of the evaluations and point out the reasons for dissatisfaction expressed in the users' own language. They can be found with a simple click, and a real-time alert system brings up the most critical verbatims...
Online Reviews - Tourism Experience
This research examines the perception of tourism. 600 opinions posted on Trip Advisor are collected according to an experience plan built on 2 destinations (Morocco, Tunisia) and 2 modes of accommodation (Hotel club and Gîtes, B&B). An initial exploration reveals that the type of accommodation has more influence on the content than the destination. The opinions are then automatically coded on the basis of a thesaurus constructed for the purposes of this research. It thus highlights that the reception experience takes precedence over the other components of the stay.
Non-directive interviews - Leadership
In this thesis, in-depth interviews conducted with 14 leaders are exploratory manner analyzed by automatic classification. It highlights 4 major refined themes in consideration of the literature on the conditions for exercising leadership. The entire interview is then read and coded according to this grid. The frequency with which the elements of the code book are mentioned in the 297 rounds of speech makes it possible to specify the importance of the dimensions of leadership and their dependence on the experience and culture of leaders. Finally a triangulation confirms the robustness of the results.
Big Data - Ecological Transition
The contributions of corpus great debate 35 000 contributions, over 10 million words, provides an overview of French concerns. The words expressions and key ideas highlighted by the lexical and semantic analysis confirm what we already knew by specifying the hierarchies and the influence of local situations. But above all, data visualization and interactivity lead to original interpretations argued by verbatim that it would otherwise have been impossible to discover in the mass of this information. Finally, the very minority comments as a percentage are found here documented by hundreds of people, which allows for deeper analysis and logic.
Bibliographic database - The employer branding relationship
This example shows how a literature review can be communicated by presenting lexical and statistical syntheses of all the articles analyzed. The major themes thus identified are then refined by targeted research corresponding to the concepts more specifically studied (thesaurus): brand / consumer relationships and the emotions involved.
The results relate to the coherence of the lines of research identified in all of the 287 articles and to their evolution over the 4 periods considered.
This example shows how the analysis of a corpus of documents can lead to interpretations based on objective data that can be checked by the reader. Data visualization makes it possible very quickly to acquire experience of the corpus studied and to discuss the interpretations proposed by the author.
Online image wall - Coberen
This example relates to the first part of a large online survey on the consumption of beverages in Europe. It involved choosing 3 images evoking the fact of drinking. The proposed image wall was made up of 25 images varying drinks, situations, motivations, etc. Respondents were then asked to explain their choice. A multilingual corpus of over 15,000 comments is thus collected.
Examination of the key words and expressions in the language of each country provides an initial idea of the answers provided.
To produce a summary, and highlight regional or country differences, an excerpt calibrated on the full base and coded online by researchers from each participating country. It can thus be established that generation and gender have more influence than the country itself, while illustrating these conclusions with verbatim in the original language.