Analyzing Twitter: All the Reactions from the Real Madrid-Atlético Match

TEXT MINING TECHNIQUES FOR ANALYZING TWITTER

Today we bring a different kind of article to the blog. By using various analytical tools, such as KNIME, we can not only explore our companies’ internal data but also look outward.

Twitter is a great resource if we want to understand our environment. Opinions, visceral criticism, praise (from time to time)… it is all concentrated on Twitter. So, taking advantage of the fact that KNIME allows for a simple connection to Twitter, we decided to put together a quick example.

THE DERBY THROUGH TWITTER

The first thing we need to create is a Twitter API. There are several sites that explain how to do this (for example, here https://www.uno-de-piera.com/crear-una-aplicacion-para-twitter/). With the Twitter API, we will obtain a series of keys and tokens, which we must copy into the KNIME connector node.

Once connected, we begin our process of extraction and information processing. In addition to the text mining tools that KNIME possesses, we design the solutions required for the ad hoc treatment of our case study (bearing in mind that many text mining tools are very well implemented for the English language, while being less useful for Spanish).

As Twitter is vast, we decided to use the Trending Topic list to analyze which hashtag would provide the most messages for analysis. That is why we used #Megaderbichampions for our study.

Thanks to various text mining tools, we can process our tweets, always with the ultimate goal of creating a Tag Cloud: a cloud of keywords from which we can extract information about our case study—in this case, the derby.

We have polarized different words for better understanding. In light of these results, it is clear that Real Madrid was the only team in the match. Specifically, Cristiano Ronaldo, who was highly discussed in the tweets (he was undoubtedly the protagonist of the encounter).

Due to the variability inherent in simply analyzing tweets, the Tag Cloud serves as a guide for deepening our understanding of the information and trying to find underlying reasons. For example, the fact that this was the hashtag used by the Atresmedia channel Mega creates a bias toward its commentators (in this case, the tag “petón” refers to José Antonio Martín Otín “Petón”, a well-known journalist and fan of the Red-and-Whites).

Two names in this cloud are particularly curious: “Messi” and “Benzema”.

WERE MESSI AND BENZEMA PLAYING?

Based on the Twitter data, we decided to investigate both concepts.

The tweets linking to Messi, contrary to what we might have initially assumed (the eternal comparison of who is better), were mainly mocking tweets, linking the fact that Barcelona had been eliminated from the Champions League to Messi showing his shirt in the last Clásico:

As for Benzema, we decided to follow the trail of the tweets and discovered something curious.

Some criticism from fans and the player’s low performance in yesterday’s match led several to consider comparisons with other forwards; could Morata have played?

THE NEED FOR AN ANALYST FOR GREATER INSIGHT

This has been a small example, conducted in an agile manner regarding yesterday’s match. KNIME allows us to carry out this text mining process on large volumes of data. We could expand the captures, the data sources, and even add new social media platforms. But even so, the supervision and “instinct” of an analyst are necessary to identify where the analysis falters, where to investigate further, and how to enrich the processing of information.