MATLAB Answers

0

How to find mutual words in title field?

Asked by bita hallajian on 6 Jul 2018
Latest activity Answered by Christopher Creutzig on 26 Nov 2018
The aim is to write a program that surveys an excel file on "title" field and distinguish the items which have mutual words in "title" field. (the excel file is attached). The question is can I use text analytics toolbox for this program if it is possible, How should I write its program? With great thanks

  2 Comments

Paolo
on 6 Jul 2018
Is it a requirement to use the text analytics toolbox? I am sure the problem can be solved easily without it too. What is the expected output?
My output is to make a graph to show the relation for the title fields which have mutual words in each line. For John & Rob, if the word "image" exists in their both title fields, there will be connection between them. The more repitition of connection cause increase the edge's weight. May I have your idea how to do? Thanks greatly

Sign in to comment.

2 Answers

Answer by Sarah Palfreyman on 6 Jul 2018

Yes, Text Analytics can help with this. It sounds like you are performing Topic Modeling, or grouping the data into sets of similar words.
https://www.mathworks.com/help/textanalytics/examples/analyze-text-data-using-topic-models.html

  0 Comments

Sign in to comment.


Answer by Christopher Creutzig on 26 Nov 2018

If I read the question correctly, you are looking for a co-occurence matrix. You can get those from a bag-of-Words model by a matrix multiplication:
txt = ["my first title", "your first idea", "my second or minute"];
td = tokenizedDocument(txt);
bow = bagOfWords(td);
cnt = bow.Counts;
full(cnt*cnt.')
ans = 3×3
3 1 1
1 3 0
1 0 4
The first row says the first sentence has three words in common with itself (duh), and one each with the second and third sentence. The second and third string do not have words in common, hence the zeroes.

  0 Comments

Sign in to comment.