"Getting started with sentiment analysis"

MikeRMikeR Member Posts: 2 Contributor I
edited June 2019 in Help
Hi,

I am seeking advice on how to get started with a sentiment-analysis in the least painfull way. I am currently writing my bachelor thesis about social behaviour on online forums. For this, i have been crawling topics on a danish forum for the last 2 months, and it finally looks like i have the data i need.

I am doing the most basic statistical analysis in SPSS, where i will compare user-rank, the amount of posts the user has made, to the amount of answers to his or her topics. However, i also have the topic text, which i would love to classify using the logic of sentiment analysis.

As you might have guessed, i am totally new to rapidminer. I have been trying to copy-paste the workflow of the accelerator sentiment analysis. But it seems i keep getting errors about my data format. However, I have only two colums: post & category. In the category column, i have mapped some of the rows with "Positive" and others with "Negative". The text in the rows is in danish, and some topics contain links, quotation marks etc.

You can have a look at my csv-file here:
https://dl.dropboxusercontent.com/u/3592722/Holdout.csv

And here's the error i get:
image

The most important two classifications i need to create/predict are:
- Positive/negative
- Subject (based on a list of subjects, with each of their keywords)

So here are the questions:

1) What am i doing wrong in the sentiment analysis?
2) Is it possible to make a prediction model, that classifies topics and labels them with subject names, based on their use of keywords(apple, win etc.)?

I have one month left to get to learn this stuff. Does that seem realistic?

Thanks in advance,

Answers

  • frasfras Member Posts: 93 Contributor II
    Hi,

    your csv file uses "," as column separator. This is not optimal because in your text column there are
    probably also a lot ",".  Choose another separator that does not exist in your text column. You may also
    use excel format where such problems do not occur. Additionally, rename your category column to
    "Sentiment".

    Cheers,
    Frank
  • abbykoenigabbykoenig Member Posts: 1 Contributor I
    Having similar issues -- I have an Excel file. One column is labeled "Sentiment" it has a few labeled Positive and some labeled Negative and the other column is labeled Text and has strings of free text. When I load it the error message says -- "For this application to work, please make sure your data contains at least one attribute of the type binominal"

    Help!

    Thank you!!!
Sign In or Register to comment.