Big data analysis and applications

• Main Author: Seow, Cheryl Marie Miao Ling
• Other Authors: Tan Yap Peng
• Format: Final Year Project
• Language: English
• Published: 2016
• Subjects: DRNTU::Engineering
• Online Access: http://hdl.handle.net/10356/67714
• Institution: Nanyang Technological University

Social media data is often overwhelming for users. Filters are used to narrow down the search results. However, due to the mass sharing culture on social media platforms, the results are still often exhaustive and endless scrolling of posts may not be helpful for the user in finding what they are searching for. Hence, this report will study the limitations of the current Instagram application and analyse datasets which are obtained from it. The report will also explain the steps taken towards implementing a recommendation system in which a combination of filters can increase the search fields pertaining to different categories. This will allow for more user inputs and accurately give users the search results which will fulfil their search terms. Instagram API does not allow developers to use location and tag endpoints at the same time. In this report, we explore possible solutions for this shortcoming. First, we use the Instagram API to get location based images and store it in an array. Subsequently, we do a search for a specific location hashtag in the json response of the images in the array. Due to the high volume of data which we expect from the results, the filtering process will be done per page with pagination url. Furthermore, the loading time of the web application is considered in the development of the recommendation system. With the use of Content Delivery Networks (CDNs) instead of the instagram url, it will reduce the page loading time as a result of reduced latency in retrieving the data.