Academify-Lite : LLM-powered social media post generator for academic content

Abstract

In recent years, social media platforms have become an integral part of many people's daily lives. What was once merely a means of communication has evolved into an essential component of businesses and job hunting. Moreover, it has emerged as a primary source of information for countless individuals worldwide. This shift in usage patterns has inspired us to leverage social media platforms as a tool for sharing scientific knowledge with the public. Our aim is to assist researchers in effectively communicating their findings to their target audience through well-crafted, understandable, grammatically correct, and appropriately toned posts. These posts aim to share the researchers' knowledge and critical findings effectively with their intended audience.

This thesis introduces Academify-Lite, a web application tool that utilizes GPT-4 as a Large Language Model (LLM). Targeting researchers, Academify-Lite allows users to input content from their academic work that they believe encapsulates the essence of their research or the key message they wish to convey. The tool then suggests multiple posts relevant to the input provided, allowing users to select the tone they believe will appeal most to their audience.

The results of this study, along with the analysis of gathered data, indicate that Academify-Lite, powered by LLM and effective prompt, significantly reduces the time researchers spend crafting posts about their research. The tool is capable of generating engaging posts that effectively convey the researchers' main message in a concise and interesting manner. This functionality streamlines the process of sharing scientific knowledge on social media platforms, enhancing researchers' ability to communicate their findings to a broader audience.