Kareem Darwish

On June 24, 2018, Turkey conducted a highly-consequential election in which the Turkish people elected their president and parliament in the first election under a new presidential system. During the election period, the Turkish people extensively shared their political opinions on Twitter. One access of polarization among the electorate was support for or opposition to the reelection of Recep Tayyip Erdogan. In this paper, we explore the polarization between the two groups on their political opinions and lifestyle, and examine whether polarization had increased in the lead up to the election. We conduct our analysis on two collected datasets covering the time periods before and during the election period that we split into pro- and anti-Erdogan groups. For the pro and anti splits of both datasets, we generate separate word embedding models, and then use the four generated models to contrast the neighborhood (in the embedding space) of the political leaders, political issues, and lifestyle choices (e.g., beverages, food, and vacation). Our analysis shows that the two groups agree on some topics, such as terrorism and organizations threatening the country, but disagree on others, such as refugees and lifestyle choices. Polarization towards party leaders is more pronounced, and polarization further increased during the election time.

In this paper, we provide a quantitative and qualitative analyses of the viral tweets related to the US presidential election. In our study, we focus on analyzing the most retweeted 50 tweets for everyday during September and October 2016. The resulting set is composed 3,050 viral tweets, and they were retweeted over 20.5 million times. We manually annotated the tweets as favorable of Trump, Clinton, or neither. Our quantitative study shows that tweets favoring Trump were usually retweeted more than pro-Clinton tweets, with the exception of a few days in September and two days in October, especially the day following the first presidential debate and following the release of the Access Hollywood tape. On two days in October 2016, pro-Trump tweet volume accounted for than 90\% of the total tweet volume.

The Paris attacks prompted a massive response on social media including Twitter. This paper explores the immediate response of English speakers on Twitter towards Middle Eastern refugees in Europe. We show that antagonism towards refugees is mostly coming from the United States and is mostly partisan.

The Paris terrorist attacks occurred on November 13, 2015 prompted a massive response on social media including Twitter, with millions of posted tweets in the first few hours after the attacks. Most of the tweets were condemning the attacks and showing support to Parisians. One of the trending debates related to the attacks concerned possible association between terrorism and Islam and Muslims in general. This created a global discussion between those attacking and those defending Islam and Muslims. In this paper, we provide quantitative and qualitative analysis of data collection we streamed from Twitter starting 7 hours after the Paris attacks and for 50 subsequent hours that are related to blaming Islam and Muslims and to defending them. We collected a set of 8.36 million tweets in this epoch consisting of tweets in many different of languages. We could identify a subset consisting of 900K tweets relating to Islam and Muslims. Using sampling methods and crowd-sourcing annotation, we managed to estimate the public response of these tweets. Our findings show that the majority of the tweets were in fact defending Muslims and absolving them from responsibility for the attacks. However, a considerable number of tweets were blaming Muslims, with most of these tweets coming from western countries such as the Netherlands, France, and the US.

Within a fairly short amount of time, the Islamic State of Iraq and Syria (ISIS) has managed to put large swaths of land in Syria and Iraq under their control. To many observers, the sheer speed at which this "state" was established was dumbfounding. To better understand the roots of this organization and its supporters we present a study using data from Twitter. We start by collecting large amounts of Arabic tweets referring to ISIS and classify them into pro-ISIS and anti-ISIS. This classification turns out to be easily done simply using the name variants used to refer to the organization: the full name and the description as "state" is associated with support, whereas abbreviations usually indicate opposition. We then "go back in time" by analyzing the historic timelines of both users supporting and opposing and look at their pre-ISIS period to gain insights into the antecedents of support. To achieve this, we build a classifier using pre-ISIS data to "predict", in retrospect, who will support or oppose the group. The key story that emerges is one of frustration with failed Arab Spring revolutions. ISIS supporters largely differ from ISIS opposition in that they refer a lot more to Arab Spring uprisings that failed. We also find temporal patterns in the support and opposition which seems to be linked to major news, such as reported territorial gains, reports on gruesome acts of violence, and reports on airstrikes and foreign intervention.

There is little doubt about whether social networks play a role in modern protests. This agreement has triggered an entire research avenue, in which social structure and content analysis have been central --but are typically exploited separately. Here, we combine these two approaches to shed light on the opinion evolution dynamics in Egypt during the summer of 2013 along two axes (Islamist/Secularist, pro/anti-military intervention). We intend to find traces of opinion changes in Egypt's population, paralleling those in the international community --which oscillated from sympathetic to condemnatory as civil clashes grew. We find little evidence of people "switching" sides, along with clear changes in volume in both pro- and anti-military camps. Our work contributes new insights into the dynamics of large protest movements, specially in the aftermath of the main events --rather unattended previously. It questions the standard narrative concerning a simplistic mapping between Secularist/pro-military and Islamist/anti-military. Finally, our conclusions provide empirical validation to sociological models regarding the behavior of individuals in conflictive contexts.

In the past several years, Arabic Information Retrieval (IR) has garnered significant attention. The main research interests have focused on retrieval of formal language, mostly in the news domain, with ad hoc retrieval, OCR document retrieval, and cross-language retrieval. The literature on other aspects of retrieval continues to be sparse or non-existent, though some of these aspects have been investigated by industry. Others aspects of Arabic retrieval that have received attention include document image retrieval, speech search, social media and web search, and filtering. However, efforts on different aspects of Arabic retrieval continue to be deficient and severely lacking behind efforts in other languages. The survey covers: 1) general properties of the Arabic language; 2) some of the aspects of Arabic that affect retrieval; 3) Arabic processing necessary for effective Arabic retrieval; 4) Arabic retrieval in public IR evaluations; 5) specialized retrieval problems, namely Arabic-English CLIR, Arabic Document Image Retrieval, Arabic Social Search, Arabic Web Search, Question Answering, Image retrieval, and Arabic Speech Search; 6) Arabic IR and NLP resources; and 7) open IR problems that require further attention.

Arabizi is Arabic text that is written using Latin characters. Arabizi is used to present both Modern Standard Arabic (MSA) or Arabic dialects. It is commonly used in informal settings such as social networking sites and is often with mixed with English. In this paper we address the problems of: identifying Arabizi in text and converting it to Arabic characters. We used word and sequence-level features to identify Arabizi that is mixed with English. We achieved an identification accuracy of 98.5%. As for conversion, we used transliteration mining with language modeling to generate equivalent Arabic text. We achieved 88.7% conversion accuracy, with roughly a third of errors being spelling and morphological variants of the forms in ground truth.