Defining text analysis models

This is the first step of the Create text analysis models wizard. In this step, you provide the information that is essential for creating a text analysis model, like the type of the analysis, language, taxonomy, and algorithms that you want to use for creating the text analysis model.

1. Expand the Analysis type list and select the type of the text analysis model.
   • Select Classification to determine whether the text units you analyze belong to any of the predefined categories.
   • Select Sentiment analysis to determine the semantic orientation and polarity in the text units you analyze.
2. Expand the Language list and select the language of the text analysis.
   • The sentiment analysis has four languages available (English, French, German, and Spanish).
   • The classification analysis has 34 languages available.

   You can add additional languages if you purchase the appropriate language license.

3. For the Classification analysis: Specify the taxonomy that you want to use in the text model analysis. The following default taxonomies are available:
   • pyBankingTaxonomy
   • pyCustomerSupportTaxonomy
   • pyTelecomTaxonomy
4. For the Sentiment analysis only: Specify the lexicon that you want to use in the text analysis model. You can use the default pySentimentLexicon.
5. Select at least one algorithm that you want to apply in the text analysis model. The following algorithms are available:
   • MAXENT - The Maximum Entropy algorithm
   • NAIVEBAYES - The Naive Bayes algorithm (available only for the Classification analysis)
   • SVM - The Support Vector Machine algorithm (available only for the Classification analysis)

   The system creates a text analysis model for each of the selected algorithms.

6. Optional: Enable spell checking in the .XLS text sample that you want to upload for machine learning purposes in the next step of the Create text analysis model wizard:
   1. Select the Enable spell checking box.
   2. Specify the spell checker property rule. You can select one of the following default spell checker property rules:
      • de - German
      • en_GB - British English
      • en_US - American English
      • es - Spanish
      • fr - French

      The spell checking functionality enables a more accurate analysis as it eliminates spelling errors from the training data.

7. Click Next.

Open topic with navigation