MeaningCloud Release: new Deep Categorization API

This is what we’ve included in MeaningCloud’s latest release:

  • New Deep Categorization API: we are happy to present the first of our Premium APIs, Deep Categorization 1.0, which lets you carry out an in-depth categorization of your data. In this initial release, we’ve included predefined models for analyzing the Voice of the Customer in several domains and the Voice of the Employee.
  • Language Identification 1.1: we say goodbye to Language Identification 1.0, so if you are still using it, you will need to migrate to the newest version. If you are using it through the Excel add-in, we’ve done it for you, so you just have to update your Excel add-in to the latest version.
  • New language for Text Clustering: we’ve added Catalan to the languages supported in the Text Clustering API.
  • General usability improvements: mainly in the developer area of the website.
New NeaningCloud release

Premium APIs and Deep Categorization

Premium APIs

We have created a new type of APIs, the Premium APIs, where we will be adding APIs with a more complex functionality and a higher added value than our basic APIs. These APIs are available for everyone: in a 30-day trial basis for users below the Professional plan, and unlimited for the rest of the plans.

Starting the 30-day trial is very easy: you just have to make a request to the API in question and the trial will begin. You will be able to check the date limit in the main page of the welcome page of the developer area, under “Products with limited access”. Requests are counted a bit differently. Each call to the API, counted as a request, is equal to the analysis of 125 words, versus the default 500 of the public/standard APIs.

Deep Categorization

The Deep Categorization API covers the need of an easier and more flexible and precise way to categorize texts. This is implemented using a very detailed rule-based language that allows to use a combination of morphological, semantic and text rules. The flexibility of this rule language enables us to create models for scenarios where precision is not only necessary but key for a good analysis.

Even though this API is new for MeaningCloud’s users, we have been using it internally for a while in different projects with many of our clients.

Two of these scenarios are Voice of the Customer and Voice of the Employee. Both are characterized by the need to analyze free text written by the customer/employee in which they can address a number of topics in varied ways.

In this first release we are publishing several models that illustrate how powerful this categorization can be:

  • Voice of the Customer:
    • Generic: for a generic domain
    • Banking: focused on the banking domain
    • Insurance: focused on the insurance domain
    • Retail: focused on the retail domain
  • Voice of the Employee, in three different use cases:
    • Performance review: for the performance reviews of your employees
    • Organization: for the employees’ opinions about the company
    • Exit interview: for the exit interviews of the employees that leave

All these models are available in English and Spanish. You can check the categories for each one of them in the corresponding documentation, or if you’d rather check an example, we have published new demos where you can test them yourself.

Other improvements

These are some of the (most visible) improvements we’ve added in this version:

  • Text Clustering 1.1: we now support a new language, Catalan.
  • Supported classification models: we have revamped the documentation for the predefined classification models, making it faster to lead and easier to browse.
  • Early restart in paying plans: some of our clients find themselves running out of credits before the billing period is over. Until now, restarting the plan was a bit cumbersome, since it forced to cancel the plan and upgrade to the same one again. To avoid this, we’ve added a “Restart” button in the subscription page where anyone will be able to restart their plan.
  • New engines + resources: we have uploaded new versions of our core engines with several bug fixes (including a false positive related to countries being considered aliases of larger entities in Topic Extraction) as well as updated our sentiment and linguistic resources.

If you have any questions or just want to talk to us, we are always available at!

Leave a Reply

Your email address will not be published. Required fields are marked *