Creating a new text model
How to start?
To create a new text model:
Click the Create New Model button.
The draft version of your model will be created automatically with a name consisting of the date and time of draft creation.
There are two scenarios for model building:
without file scenario: Build your model from scratch. This option is already selected at the start.
file scenario: Use data from a CSV file to build your queries and models.
You can switch between scenarios at any time by clicking the gray button on the top left. However, please note that your progress will not be saved unless you generate your model.
Without file model building
In this section, you can create your query using words, phrases, or predefined clusters.
Configure your query by providing a Topic / label (concept) in the text field.
It can be a single word, topic, phrase, or value that your text model has to extract. For example, you can use a sentence like "red car has engine problem", a short phrase like "drugs availability" or a single word like "emotions".
Click the Add context button, if you want to add information that specifies the concept.
You can build concepts using words written together (e.g., query "car problem" will identify all car-related problems) or separately (e.g., "car" + "problem") to identify information related to cars and/or problems.
Expanding your query
You can use clusters to expand your query:
Click on the vial icon to search for aggregated concepts associated with the word you entered. There are two search options:
Contains: Filters words that contain the word you entered, e.g., highlights words like "card" or "care" for the word "car".
Whole Word: Filters words that match the whole word, e.g., highlights words like "car" or "car seat" for the word "car".
Predefined clusters are available in various topical domains. Double-click to add them to your topic / label or context section.
Building your query
To build a query, there are two options:
SYNONYMS: Get a query for concepts with the same meaning (synonymous).
SIMILARS: Extend your queries with similar terms but in a much wider context.
You can configure and build as many queries as you want.
Annotating text
For the entered text, select the desired annotation type to identify specific linguistic elements.
example sentence: The little girl wandered through the beautiful garden, pausing occasionally to smell the fragrant flowers.
All: Annotates all phrases.
Noun Phrase: Labels noun phrases, which include a noun and its modifiers.
Verb Phrase: Labels verb phrases, which consist of a verb and any object or modifier directly related to it.
Descriptive Phrase: Labels descriptive phrases that provide additional information about a noun or a verb.
With file model building
This section will be updated with step by step guide shortly.
Queries preview and model generation
In the MY QUERIES section, you can review your built queries, preview matched clusters' content, and modify them by removing words/groups from the base and context fields.
To generate your text model, select the query(ies) you want to use and click the GENERATE YOUR MODEL button. You will be redirected to the list of your private text models, where you can activate your model for use in analysis.
Last updated
