Big Data and Trust in Social Media Analytics

Big Data and trust in Social Media Analytics

Mario Viso

Abstract—Big data analytics is a great tool for extracting meaning from large amounts of data. As the connection with social media networks grows, they have become  an  ideal  source for gathering data in different formats and providing unique benefits. If the whole procedure is ethically and transparently done, then we can gain a lot of profit and social media analysis can have many benefits. However, trust in social media has been questioned in the past so it must be guaranteed that there is legal compliance concerning data privacy issues such as misuse and manipulation of data. Our lives can change for the better if the right data is used in the right way and for the right reasons, especially if social media analysis is used transparently.

Keywords—Big data, Analytics, Privacy, Social media analytics, Social network sites, Predictive analytics, Social media trust, transparency

1. INTRODUCTION

As organizations make more use of big data, it is likely that there will be increased concerns and legislation matters about individual privacy issues. The present work focuses on the relation of social media analytics with big data and examines the level of user’s trust in the social media field. First, all the theoretical background needed from a technical perspective about the data analysis procedure is introduced. The notion of

vagueness is added as a 4th V of Big Data which is useful in the social media context. Furthermore, different kinds of analytics used in the data science phase are mentioned.

Recommender Systems and Collaborative Filtering are explained as they are more targeted tools used for social networks analysis. Advantages and disadvantages of using Big data analytics are discussed in order to weigh both and have a clearer opinion on the value and benefits of the results if the whole procedure is transparently done. Furthermore, user’s trust in social media world is discussed. The way data is stored, accessed and used by different actors raises many moral issues and risks. Trustworthiness of information as well as measures of user’s trust in social media by specific algorithms are addressed. In addition, we see how user’s behavior and decision making is impacted by the level of his trust. Moreover, a possible real-life example is demonstrated on how an individual’s choice on voting could be affected positively or negatively. Last but not least, some governmental issues concerning access in personal data are discussed.

1. THEORETICAL BACKGROUND

1. Big data characteristics

Big data is normally used to describe the fast growth and availability of any kind of data. It can be applied to find out

[pic 2]

Fig. 1.  Illustration of Social Big Data

the relationship and dependencies among data or datasets or even to perform predictions of outcomes and behaviors of the existing data set. Social media is one of the most important and significant sources of big data. [1] Usually with social media, the data will come at a high speed and in a variety of format such as text, audio, video, pictures. In order to, extract knowledge, manage, organize and give sense to this data only big data applications or big data analytics can be used.

Big data is defined as data that requires at least one of the following dimensions or so called 4 Vs: volume, variety, velocity and veracity. In the social media context these characteristics would be:

• Volume: Social media data are large in size

• Variety: Data can be structured, unstructured or semi- structured. It can be tweets which are texts, Wiki media which can be XML (semi-structured data), and Facebook data which can be photos, videos in addition to texts. Furthermore, how users are connected between social media can be represented by graphs (networks)

• Velocity: The speed at which data grows is very high

e.g. tweets.

• Vagueness: Instead of veracity the characteristic of vagueness was proposed as a 4th V for social data by Venkat Krishnamurthy on Big Data Innovation Summit in Silicon Valley in 2014, which refers to the confusion over the meaning of Big Data. [2] [3] [4]

According to Ishikawa, the characteristic of vagueness is a result of a combination of various types of data to be analyzed, which lead to inconsistency and deficiency. [5] It also relates to the issues of privacy and data management as social data involve individuals’ personal information.

Big data analytics can be divided into descriptive (describe data), exploratory (discovering unknown correlations in data), predictive (predict events and trends) and prescriptive (suggest actions) analytics. [6] [7]. Social Network Analysis is one of the most established fields of data analytics [7], providing tools, methods and theories for the research of social networks in the digital realm.

1. Recommender Systems and Collaborative Filtering

Besides all foundational background some more targeted tools are shown below. Firstly, Recommender Systems are tools that use user preferences, opinions, and predict user’s interests so that they can suggest and show them targeted items [8]. Moreover, communities’ detection algorithms group users with similar interests and similar nature of interactions to improve the sparsity issues and efficiency of recommendation algorithms [9] [10].

Collaborative filtering Recommender Systems acquire user ratings on products or product features and learn the user interests from ratings to personalize the product. [11] Collaborative filtering uses the ratings of untold numbers of users to find similarity user-user or item-item to personalize the items for a given user [12]. A Collaborative filtering can  be based on a training model for generating recommendations. The model-based Collaborative filtering improves the sparsity issues and computational efficiency by using reduced features instead of the whole dataset. [13]

1. Advantages and disadvantages of Social media analytics

The benefits we get from social media analysis cannot be overseen especially as social media analytics research field is getting bigger and easier to understand thanks to frameworks like fig. 2. [14]. Social media and big data nowadays can accelerate the process of innovation. [15]. At Google, social media, as well as the 20% rule dedicated to autonomous projects, led to product and service innovations, including the

[pic 3]

Fig. 2. The Social Media Analytics Framework together with challenges in its context