How Representative Are Social Media Datasets of the True Population: A Case for London

Introduction Name research provides valuable insight into human structure and dynamics (Longley et al., 2011). Individual names are considered to well reflect the cultural, ethnic and linguistic characteristics of the owner (Mateos et al., 2011). The usefulness of this information unique to individual names led to the development of the world name database (WND: http: //worldnames.publicprofiler.org/); comprehensive census and telephone directory for the 26 continents of the four continents data. It is about 2 billion people in the Earth's population.

The concept of "prejudice" is understood in several ways. In statistics, a dataset or sample is considered to be skewed if the dataset or sample is systematically different from its target population. As in everyday language, in ethics, if a decision fails to treat people fairly, the decision is usually considered to be biased. In either case, because prejudice includes partial or unilateral insight, people will make misunderstandings. Algorithm bias occurs for several reasons. First, the data used to train the machine learning model is often incomplete or distorted. By representing or excluding certain marginalized groups or groups of societies, this "sampling error" leads to deterioration of products with insufficient calibration, rather than offsetting the stuff left out.

Weighting, also known as sample balance, is a technique used to reflect differences in the number of populations represented in each case in the data set. Normally, in a study designed to represent the population of the United States, the units are adjusted to reflect the US census. Various methods can be used in the searching agency's weighting process, but usually with weighting, when analyzing data, you multiply the survey results by one or more factors to increase or decrease the importance of the observation value.

Social media is the largest dataset of unstructured human expression that has been recorded so far and is growing exponentially today. According to Gary King, a professor at Harvard University and Director of the Institute for Social Science Numbers, one billion social media are posted every two days. Listening to social media is increasingly interested to product developers and marketers, but in Australia in particular the industry has not completely accepted the use of social data for business intelligence. Given that finding conversations related to your business may be like sifting a single sand to find the salt you are throwing away, this is not surprising