While some news anchors try to stay professional and subjective in all of their articles, most of the news we consume are published to push a specific agenda especially when it comes to politics.
In our effort to battle news bias, it is crucial for our system at Almeta to understand this hidden Agenda, in this article we will explore one path to do this.
This article is a part of our series on political bias detection we will hopefully introduce you to the various aspects of our political bias detection system, and you can learn about:
- How can we predict the political orientation behind a piece of the news?
- What is Stance detection? and what are the different types of it?
- What is subjective stance detection? what is distance supervision? and why they make a cute couple?
- What is ALSA, ELSA and how can opinion mining save us from political bias?
- How to implement an initial political bias detector just from sentiment analysis and some probabilistic distribution? (warning cool visualizations)
- How to Visualize a Political Bias Metric
Let us start with a very crucial assumption, we assume that if an article is extremely in favor (or against) a certain entity or idea then this article is biased. This is the simplest axiom we can start our discussion from.
Although this task does not strictly fit within the boundaries of the opinion mining family it fits the assumption we had, here the entity is the political ideology, in this task, given an article, the system should predict the political orientation of that article (in favor of which ideology is it).
In western politics, the political spectrum is more uniform with a clear dichotomy between the left and right, in that case, the system boils down to a binary classifier. However, in Arabic politics, the vision is more blurred with various orientations.
The How (The Data)
For the western politics the congressional and parliament debates make for a great source of data since by simply knowing the name of the representative it is possible to know his/her political orientation. For example (democrats usually hold liberate left ideologies while republicans have a conservative right ideology) this dichotomy helps
However, this type of data collection does not work in Arabic politics since no such dichotomy exists. Since the role of the party is rather
The only work we found working on this in Arabic was  here the authors consider only 5 types of political ideology present in the Arabic politics and scraped data for them from news sites that hold the same ideology, the authors focused on articles that talked about a certain news in order to consider the way the same news is seen by the different ideological sites, while discarding any articles that directly talks about the ideology and its history. The extracted data is here the reported size of the data set is fine yet we didn’t assess its quality, and in case more data is needed, applying the method outlined in the paper is straight forward (either by using the same sites or similar ones), the following table shows the ideologies considered in the dataset and the source sites chosen by the authors.
|Arab Nationalism||http://www.alfikralarabi.org/index.php http://www.qawmi.com/|
|Islamic Brotherhood||This http://www.ikhwanonline.com/ http://www.dd-sunnah.net/ https://www.egyptwindow.net/|
The How (Code)
The task boils down to simple text classification, and while the authors in  used feature engineering, a simple neural model
Hopefully by now you should have an essential Idea of what is the political orientation and how it is possible to distinguish between different orientations of different text pieces. and as always don’t forget to check the resources.
Do you know that we use all this and other AI technologies in our app? Look at what you’re reading now applied in action. Try our Almeta News app. You can download it from Google Play or Apple’s App Store.
 M. Iyyer, P. Enns, J. Boyd-Graber, and P. Resnik, “Political ideology detection using recursive neural networks,” in Proceedings of the 52nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), 2014, vol. 1, pp. 1113–1122.
 R. Abooraig, S. Al-Zu’bi, T. Kanan, B. Hawashin, M. Al Ayoub, and I. Hmeidi, “Automatic categorization of Arabic articles based on their political orientation,” Digit. Investig., vol. 25, pp. 24–41, 2018.