Term Document Matrix

In a Term-Document Matrix (TDM) or Document-Term Matrix (DTM):

• Rows represent terms (words or sometimes phrases).
• Columns represent documents.
• Entries in the matrix indicate the frequency (or sometimes weighted frequency) of a term in a particular document.

For example, consider you have the following three documents:

1. "I love apples."
2. "Apples are sweet."
3. "I love sweet food."

A simple Term-Document Matrix might look something like:

	Doc1	Doc2	Doc3
I	1	0	1
love	1	0	1
apples	1	1	0
are	0	1	0
sweet	0	1	1
food	0	0	1

In this matrix:

• A row represents a term.
• A column represents a document.
• The value at a given cell (term, document) represents the frequency of that term in the document.

A common enhancement to raw frequencies in a TDM is the use of TF-IDF (Term-Frequency-Inverse-Document-Frequency). TF-IDF weighs terms by their importance, giving higher weights to terms that are frequent in a specific document but not across many documents in the corpus.