Predicting X from Twitter is a popular fad within the Twitter research subculture. It seems both appealing and relatively easy. Among such kind of studies, electoral prediction is maybe the most attractive, and at this moment there is a growing body of literature on such a topic. This is not only an interesting research problem but, above all, it is extremely difficult. However, most of the authors seem to be more interested in claiming positive results than in providing sound and reproducible methods. It is also especially worrisome that many recent papers seem to only acknowledge those studies supporting the idea of Twitter predicting elections, instead of conducting a balanced literature review showing both sides of the matter. After reading many of such papers I have decided to write such a survey myself. Hence, in this paper, every study relevant to the matter of electoral prediction using social media is commented. From this review it can be concluded that the predictive power of Twitter regarding elections has been greatly exaggerated, and that hard research problems still lie ahead.

/pdf/i-wanted-to-predict-elections-with-twitter-and-all-i-got-was-4eui4bye6d.pdf

"I Wanted to Predict Elections with Twitter and all I got was this Lousy Paper" A Balanced Survey on Election Prediction using Twitter Data

Product sales forecasting using online reviews and historical sales data: A method combining the Bass model and sentiment analysis

Social media platforms have democratized the process of web content creation allowing mere consumers to become creators and distributors of content. But this has also contributed to an explosive growth of information and has intensified the online competition for users attention, since only a small number of items become popular while the rest remain unknown. Understanding what makes one item more popular than another, observing its popularity dynamics, and being able to predict its popularity has thus attracted a lot of interest in the past few years. Predicting the popularity of web content is useful in many areas such as network dimensioning (e.g., caching and replication), online marketing (e.g., recommendation systems and media advertising), or real-world outcome prediction (e.g., economical trends). In this survey, we review the current findings on web content popularity prediction. We describe the different popularity prediction models, present the features that have shown good predictive capabilities, and reveal factors known to influence web content popularity.

/pdf/a-survey-on-predicting-the-popularity-of-web-content-2z4u0a0s0q.pdf

A survey on predicting the popularity of web content

As the age of Big Data emerges, it becomes necessary to take the five dimensions of Big Data - volume, variety, velocity, volatility and veracity - and focus these dimensions towards one critical emphasis - value.The Encyclopedia of Business Analytics and Optimization confronts the challenges of information retrieval in the age of Big Data by exploring recent advances in the areas of knowledge management, data visualisation, interdisciplinary communication and others. Through its critical approach and practical application, this book will be a must-have reference for any professional, leader, analyst or manager interested in making the most of the knowledge resources at their disposal.

Encyclopedia of Business Analytics and Optimization

First a very simple but important point about statistics. Statistics cannot be used to establish causation, they can only show correlations. As an example, the governor of a certain state is concerned about the test scores of high school students in the state. One of his aides brings up an interesting statistic: there is a very strong link between student test scores and the taxes paid by the parents of the student. The parents of high scoring students pay more taxes. The aide’s suggestion for increasing student test scores is unusual; sharply increase tax rates. Surely student test scores will follow! The fallacy, of course, is that even though there is a correlation between test scores and parents taxes, there is likely no causal connection. A better explanation is that there is some hidden variable that explains the correlation. In this case the obvious choice is the income of the parents. This determines the taxes paid. And since the quality of high school that a student attends is to a large extent determined by the parent’s income, we see a causal link from parent’s income to both taxes and test scores.

How to lie With Statistics

Social media comprises interactive applications and platforms for creating, sharing and exchange of user-generated contents The past ten years have brought huge growth in social media, especially online social networking services, and it is changing our ways to organize and communicate It aggregates opinions and feelings of diverse groups of people at low cost Mining the attributes and contents of social media gives us an opportunity to discover social structure characteristics, analyze action patterns qualitatively and quantitatively, and sometimes the ability to predict future human related events In this paper, we firstly discuss the realms which can be predicted with current social media, then overview available predictors and techniques of prediction, and finally discuss challenges and possible future directions

A Survey of Prediction Using Social Media

With the evolution of the software development, the scale of the software is increasingly growing to the extent that we cannot hand it easily. Some metrics are proposed to measure the complexity of software in last a few years. This article aims at a comprehensive survey of the metric of software complexity. Some classic and efficient software complexity metrics, such as Lines of Codes (LOC), Halstead Complexity Metric (HCM) and Cyclomatic Complexity Metric (CCM), are discussed and analyzed first. Then, some other approaches driven from above classic metrics are also discussed. The comparison and the relationship of these metrics of software complexity are also presented.

A survey on metric of software complexity

In order to make lots of new malwares fast and cheaply, attacker can simply modify the existing malwares based on their binary files to produce new ones, malware variants. Malware variants refer to all the new malwares manually or automatically produced from any existing malware. However, such simple approach to produce malwares can change signatures of the original malware so that the new malware variants can confuse and bypass most of popular signature-based anti-malware tools. In this paper we propose a novel byte frequency based detecting model (BFBDM) to deal with the malware variants identification issue. The byte frequency of software refers to the frequency of the different unsigned bytes in the corresponding binary file. In order to implement BFBDM, two metrics, the distance and the similarity between the suspicious software and base sample, a known malware, are defined and calculated. According to the experimental results, we found out that if the distance is low and the similarity is high, the suspicious software is a variant of the selected malware with very high probability. The primary experimental results show that our model is efficient and effective for the identification of malware variants, especially for the manual variant.

Detecting Malware Variants by Byte Frequency

Fast-flux refers to rapidly changing the mapping between IP address and domain name. Although some benign uses with this technique are known, it currently has become a favorite tool for cyber criminals to launch collaborative attacks, such as phishing, pharming, and malware spreading. While the legal fast-flux networks and the malicious ones hold some same features, such as short TTL and large IP pool, it is hard to distinguish them. In this paper we propose a novel way to deal with the fast-flux attack identification issue. We try to measure the service availability of the agents in the fast-flux network to identify the malicious fast-flux. This is the first time that researchers observe the fast-flux network in terms of service availability. We develop some metrics on the service availability. And the observation results show the metrics are useful.

Fast-flux attack network identification based on agent lifespan

Within social networking services, users construct their personal social networks by creating asymmetric or symmetric social links. They usually follow friends and selected famous entities, such as celebrities and news agencies. On such platforms, attention is used as currency to consume the information. In this chapter, we investigate how users follow famous entities. We analyze the static and dynamical data within a large social networking service with a manually classified set of famous entities. The results show that the in-degree of famous entities does not fit to a power-law distribution. Conversely, the maximum number of famous followees in one category for each user shows a power-law property. Finally, in an attention economics perspective, we discuss the reasons underlying these phenomena. These findings might be helpful in microblogging marketing and user classification.

Sheng Yu

Papers

A Survey of Prediction Using Social Media

A survey on metric of software complexity

Detecting Malware Variants by Byte Frequency

Fast-flux attack network identification based on agent lifespan

Social Network Dynamics: An Attention Economics Perspective