How is the connection between Text Mining, NLP and Tasks like Tokenization, Lemmatization, Stop-word Removal etc.?

Question

I am new to the whole world around Big Data and Text Mining.

It took me a while to understand all the connections and to be able to classify the buzzwords.

But there's one thing I still don't understand. The connection between NLP, text mining and tasks like tokenization, lemmatization, stop-word removal etc..

I refer to these two papers for example:

• https://www.elderresearch.com/wp-content/uploads/2020/10/Whitepaper_The_Seven_Practice_Areas_of_Text_Analytics_Chapter_2_Excerpt.pdf
• https://www.researchgate.net/publication/311394659_Text_Mining_Techniques_Applications_and_Issues

how do I connect this?

Option 1:

• Tasks like tokenization, lemmatization etc. are tasks of NLP
• and NLP is an application field of text mining

Option 2:

• Tasks like tokenization, lemmatization etc. are tasks of Text Mining
• which find their usage in NLP?

Can someone explain this to me?

Answer 1

From a research perspective, the domain is called Natural Language Processing (NLP). This is the term that people use to describe their specialty, to name their teams, big conferences, etc. For the sake of completeness I have to mention that the term Computational Linguistics is also used quite a lot (sorry for adding another term to your confusion!).

To my knowledge the term "text mining" is never used as a scientific field, and actually not used that much. Wikipedia defines text mining as the general process of deriving information from text, so from this point of view it's a general term which includes most of NLP. So technically your option 1 is probably the most correct, however I don't think anybody would ever say that "NLP is an application field of text mining", it doesn't sound right because text mining is not the name of the domain.

In the usage "text mining" usually refers to the exploratory (often unsupervised) side of applications, in a way which is somewhat similar to what data mining is to machine learning. But honestly I don't think it's worth the effort of trying to define formally the relationship or exact limits of these concepts which overlap a lot and evolve quite fast anyway. In other words: don't overthink this ;)

For the record NLP is in at the intersection of many fields or subfields: it overlaps with speech processing, information retrieval, knowledge representation, data mining, etc.

Anyway, welcome to the field :)