How to use "text mining" tool which blatantly understand what you usually say on Twitter



For a collection of sentences, analyze how frequently words and phrases appear, what kind of correlation, what kind of time series are appearing and get useful information "Text miningUser local release tool which can do easily. The analysis target is a text file and a Twitter account, you can see what kind of remarks you usually do.

Text Mining Free by User Local
http://textmining.userlocal.jp/


When using it, access to the site and clicking on each "analysis page" OK.


For Twitter analysis, account linkage is required, so enter ID and password and click "Login".


I also tried it with GIGAZINE official account. In the "co-occurrence network" in which the appearance patterns of words appearing in sentences are similar, we can see that there are many occurrences such as "eat", "use", "possible" etc. "Movie" has overwhelming presence in the word cloud.


A score by noun, verb, adjective is like this. From the frequency of occurrence and the words that are coming out, it seems that the tweets are using something weak for the last two weeks.


One text file analysis is applicable to text files (.txt) and CSV file (.csv) files whose character codes are UTF-8 · Shift_JIS and can analyze up to 100,000 characters. Simply select the analyzed file and press "analyze".


"Aozora Bunko"Running Meros(Author: Osamu Dazai), and this is the result. The number of occurrences of "Melos" is large, and it is understood that the connection with nouns, verbs, and adjectives appearing in the work is also strong.


Even in the word cloud, obviously the center is Meros. Selinuntius and tyrant are about the same size.


As I concretely look, the frequency of occurrence of "Melos" is overwhelming though I knew it. In verbs, it seems that the frequency of appearance of "coming" is higher than "running".

in Review,   Web Service, Posted by logc_nt