Most frequent word in text

str = '123 zxy abc def abc def abc last word';
% str = fileread(fileName); % Read in text from disk file.
words = strsplit(str);
uniqueWords = unique(words)
numUniqueWords = length(uniqueWords)
wordCounts = zeros(numUniqueWords, 1);
for k = 1 : numUniqueWords
	thisWord = uniqueWords(k);
	indexes = ismember(words, thisWord);
	wordCounts(k) = sum(indexes);
end
% Show results in command window
wordCounts

Do you have the Text Analytics Toolbox? There are probably functions in that toolbox to get a histogram of words easier than this.

3 个评论
显示 1更早的评论隐藏 1更早的评论

Walter Roberson 2019-11-27

在 MATLAB Online 中打开

If you have a cell array of character vectors that is the words, then you can use

randperm(number_of_words, number_to_choose_randomly)

to get out a cell array of character vectors that are that many randomly choosen words. After that your task is reduced to one of displaying them, such as

fprintf('%s\n', TheCellArray{:});

Image Analyst 2019-11-27

Roger, you might find Talk To Transformer fun. It will generate sentences using a neural network. So even though it generates gibberish, it's not just random words. The grammar is right with nouns, adjectives, etc. right, and sentence structure right. For example when I type in "I like to use MATLAB Answers." below is how it completed the paragraph.

"I like to use MATLAB Answers. There's one new way to run a simulation if I have time, and that's to run the Model of a Power Grapher experiment with a mesh that's made of a grid that covers the corresponding coordinates. The reason for this is that the Lattice Proximal layer doesn't cover each coordinate perfectly, meaning that each layer overlaps some areas, which introduces a kind of noise to the output image. My current theory is that the noise causes the software not to converge as well. Unfortunately, I don't have the equipment."

I've seen one professor feed the whole works of Shakespeare into a network and after the first epoch it was just random letters, than after a few hundred more, it was breaking them into words, then sentences. And after even more it was getting grammar right. After more and more epochs the text got more and more reasonable and less gibberish sounding. He thinks if he trained it for weeks, it might produce something that sounded very reasonable.