Question regrading to the method of news sentiment score

Hi all,

I am wondering what method the news sentiment analysis uses. I tried both textblob and nltk python library to make sentiment score for news, but they got totally different results comparing with the index shown on the right of the news in News monitor (that index is relatively correct).

For example the news with storyId 'urn:newsml:reuters.com:20211122:nL1N2SD2VH:3' got both positive scores with textblob and nltk library, however the news page shows it is negative.

So how can I get the same result with that index? Thanks in advance

Find more posts tagged with

news

python api

sentiment

Accepted answers

raksina.samasiri

hi @xiaodong.han

Based on the answer from this thread

I've just confirmed with the Eikon help desk that The Eikon Data API supports News Headline and News Story only. However, the application can use other Python libraries to analyze news sentiment from the headline/story. Please see more detail in Introduction to News Sentiment Analysis with Eikon Data APIs - a Python example article.

All comments

raksina.samasiri

hi @xiaodong.han

Based on the answer from this thread

xiaodong.han

Hi raksina.samasiri ,

Thanks for your answer. I have read those links already, but it did not solve my problem. The sentiment score shown in the news seems not to calculate neither by nltk nor textblob. Is it calculated by other libraries or other methods? That is my question.

raksina.samasiri

Hi @xiaodong.han

This forum is dedicated to software developers using Refinitiv APIs. The moderators on this forum do not have deep expertise in every bit of content available through Refinitiv products, which is required to answer content questions such as this one.

The best resource for the content questions is the Eikon support team, which can be reached by submitting queries through MyRefinitv. The support team will either have the required content expertise ready available or can reach out to relevant content experts to get the answer for you.

However, in this case, I open case number 10755610 on behalf of you and the support team is going to contact you soon.

xiaodong.han

Thanks for your help sincerely! Someone has reached me via e-mail.

tesfaye.salarin

Hi! I'm new to the community and was wondering the same question for my research, could you share more info about it?

EXPLORE OUR SITES