T-LAB Plus 20186 January 2018
T-LAB Plus 2019
was released on October 13th 2018
Here is a short list of the most significant improvements made in this version of the software.
1 - the Concordances tool now includes a new option which allows the user to build/explore a dynamic Word-Tree of any selected item (see below). N.B.: The Google chart library is used for this purpose.
2 - the way the Corpus Builder tool manages the CSV files in different languages has been significantly improved. Moreover, whenever the dataset includes texts consisting of up to 30,000 characters each (i.e. about 10 pages each), a CSV backup copy of the corpus is made automatically available which can be edited using any spreadsheet editor and also re-imported into T-LAB.
3 - the Modeling of Emerging Themes tool, which uses the Probabilistic Topic Model (Blei, 2012; Steyvers & Griffiths, 2007), has been completely redesigned. As a result, the user has three main advantages:
a) they can analyse larger data matrices (i.e. sparse matrices including up to 5 million word occurrences) without any restriction on the number of documents contained in the dataset;
b) they can use the default options or customize the various analysis parameters;
c) they can explore the analysis results through lots of customizable tables and graphs, either for exploring the relationships between the various themes (i.e. topics) or for exploring the relationships between words within each theme (see the below pictures for some examples).
Moreover, in order to improve the quality of the results and to increase the integration of various analyses, the context units (i.e. the rows of input matrices) are always elementary contexts, like in all T-LAB tools dealing with word co-occurrences.
All other options of this tool (e.g. rename/remove items, assess the semantic quality of each thematic cluster, classify documents etc. ) remain unchanged.
4 - the Graph Maker tool, which is available in T-LAB whenever the user is dealing with word co-occurrences and word clusters (including the above listed Modeling of Emerging Themes), now allows us to build/customize nine different types of graphs and to export files in several formats which can be imported into other software programs for further analyses.
5 - The comparative analysis sub-menu now includes a new tool for performing a Singular Value Decomposition (SVD) of various matrices with up to 300,000 rows and 5,000 columns. By using such a tool it is possible to extract the first 20 latent dimensions of the analysed matrices (including co-occurrence matrices) and to display the results by means of charts and tables (see below). Moreover, by using this tool, it is possible to explore similarity measures based on word embeddings.
6 - the Correspondence Analysis tool now includes two new options which allow the user to check the specificity measures (i.e. chi-square and test value) of each word listed in the contingency table.
7 - the Word Associations tool, besides providing measurements (i.e. indices) and charts which allow the user to explore one-to-one relationships between single words, now provides a way of exploring the co-occurrence contexts of any word pair.
8 - the Variable Manager tool now includes a new button which allows us to plot charts both with the occurrences of single variables and with crossed variables.
9 - All Analysis tools now include some hints at the top of the corresponding window, which should be helpful for inexperienced users (see the below image).