Skip to content
  • There are no suggestions because the search field is empty.

Understanding Natural Language Processing (NLP)

Natural Language Processing (NLP) is a branch of artificial intelligence that deals with the interaction between computers and humans using natural language.

Datamaran's engine applies different NLP techniques. The engine interprets the narrative in a document by analyzing the relationship between the key terms and concepts mentioned.

Our data scientists have used NLP techniques to develop algorithmsto match a manual analysis. This is what we call "automating human expertise." It continuously compares the results from the human and the machine analysis to ensure consistency and robustness in the engine's results.

See below some examples of how Datamaran applies NPL.

AI algorithms
Datamaran source
How is Datamaran using it?
Why is it important?
Example
Language detection All Identifying which content to include in the analysis Algorithms are usually set up to work with 1 language The body of text fed to the machine needs to be "sorted" to apply the correct algorithm corresponding to the appropriate language
Organization detection All Locating mentions of companies in a body of text The engine can understand if the context refers to a company or not, in order to assign a score to it Several company names are also commonly-used words (e.g. Apple, Amazon, Orange…) that need to be differentiated
Information retrieval Corporate reports, regulations Understanding the level of emphasis put on a topic Allows the engine to refine the assessment of topics analyzed The tool identifies the different ways in which a topic is covered or mentioned in reports and how to identify the most important source (e.g. CEO letter vs. footnote)
Fake news detection News Focusing analysis on reliable, trustworthy sources It increases the quality of results that are otherwise impacted by false statements Fake news usually follows a different pattern to "real news." This pattern can be detected and articles removed from the database used for the analysis