A Dataiku DSS recipe is a set of actions to perform on one or more input datasets, resulting in one or more output datasets. The MeaningCloud plugin provides a number of recipes with the following characteristics:

  • They receive a single dataset as input with some unstructured data to analyze.
  • The unstructured data is sent to the MeaningCloud APIs to obtain the analyses requested.
  • They output a new dataset with the original data plus the analyses done.

The MeaningCloud plugin contains five recipes:

Language Detection Detect the dominant language of a text (language name and ISO 639 code) using MeaningCloud's Language Identification API.
Sentiment Analysis Analyze the sentiment polarity, subjectivity, irony and emotional agreement of a text using MeaningCloud's Sentiment Analysis API.
Topic Extraction Extract Named Entities (people, organizations, etc.), concepts, money expressions and quantities from a text using MeaningCloud's Topics Extraction API.
Text Summarization Summarize a text according to a specified number of sentences using MeaningCloud's Summarization API.
Deep Categorization Assign one or more categories to a text using advanced rule-based language models using MeaningCloud's Deep Categorization API.

The following image contains an example of the recipes provided by the MeaningCloud plugin used with some of the datasets we've previously shared in the tutorials in our blog.

Example flow with recipes

In the following section, we will see in detail the different analyses the recipes in the plugin provide.