logo © 2023 DBS@Heidelberg
faq
FAQ
What is CQE?
CQE stands for comprehensive quantity extractor, which is a tool for extracting quantities and relative concepts from unstructured text. Each quantity comprises a value, unit, change, and concept that it refers to. For example, in the sentence "German DAX fell by 2 percent", the value is 2, the unit surface form is percent and its normalized form is a percentage. The stock index is falling and therefore has a downward trend or change and the value and unit refer to the concept German DAX.
How to use the system?
Simply enter the desired text in the textarea, copy and paste it, or simply drag and drop any .txt file. By clicking on the Extract button annotations are updated. You can insert and compare the extraction of multiple documents by clicking the + button and inserting a new text.
How can I download the annotations?
After extraction, annotations are available for download in JSON format by clicking the download icon beside the extract button.
Alternatively, you can use the API endpoint /api/extract to retrieve the same information programmatically. For this, wrap the text in a simple JSON structure and send it as the payload of a POST request:
curl https://quantplorer.ifi.uni-heidelberg.de/api/extract \
    -X POST \
    -H "Content-Type: application/json" \
    -d '{"text": "My grandmas recipe calls for 100g of butter." }'
Can I use CQE in my project?
Yes, CQE is also available as a pip package. You can install it using pip install CQE or by visiting the GitHub repository: .
How to filter for specific quantities?
By clicking the Filters tab, you can filter results based on distinct units, ranges of values, specific changes, and concepts. After filtering only the remaining quantity annotations are exported through the download button.
How to use the statistics tab?
The statistics tab provides three kinds of statistics to provide an overview of the annotations.
Unit Distribution shows all units present in the document and the frequency of their usage.
Concept Distribution, similarly counts the occurrence of distinct concepts.
Value Distribution is more elaborate and allows for an understanding of the distribution of values for a specific concept and unit. For example, one can explore how a specific stock market value changes in the document based on percentage or dollar.