Impactfulness – for solving a real-world problem
As specified in the problem statement, automatic text summarization for different types of sources is an extremely useful tool aiding people in browsing efficiently through the immense sea of information. The importance of text summarization lies in its ability to capture essential ideas that consolidate readers’ understanding of the texts in an effective and efficient approach.
The aim of this Hackathon is to extend the capabilities of Ekatra, a low-data learning platform providing education to underprivileged students without stable access to high-speed internet, which is especially relevant during the pandemic where remote learning prevails. Using text summarization can further minimize the amount of data needed by allowing students to read the summaries and determine whether the sources are useful and needed, before spending data to access the complete documents. As a result, developing the text summarizer will not only help the 75% of students in India impacted by the lockdown, but also students all around the globe facing similar troubles brought by unfamiliarity with technology, lack of resources, and other concerns in the digital learning era.
As COVID-19 continues to bother societies around the globe, and as long-distance learning turns into a normalized phenomenon, our text summarizer and its added-on features provide an implementable solution to this trending global issue. Moreover, it is not only useful for solving problems associated with the Ekatra project and the Indian society but also potentially helpful on a larger and wider scale, whenever similar barriers to education persist in the real world.
Innovativeness – for use of GPT-3
We incorporated GPT-3 in the text summarizer we developed. Ever since its initial release in 2020, GPT-3 excites the users with its inestimable competency. It earned wide recognition and was often praised as the most powerful language model.
Different from other existing summarizing algorithms and models, GPT-3’s ability to work with minimal input and maximal parameters makes it the ideal model to use. Providing only instructions, we can expect GPT-3 to generate text summarization that is accurate, comprehensible, and suitable to humans’ needs.
In addition, GPT-3 is so powerful due to how much it can accomplish. Besides plain text summarization, GPT-3 is also capable of outstandingly completing a number of diversified tasks. Thus, we tried to decorate our solution with creativity, by adding features supported by GPT-3 that are relevant to the challenge prompt. For example, since GPT-3 supports multiple languages, we added a translation feature, allowing users to obtain summaries of sources not originally produced in their native languages. We also added a length feature so that users will be able to customize summaries possessing different levels of precision and amounts of detail according to their needs. We not only used GPT-3 in coming up with a text summarizer but also capitalized on GPT-3’s diverse range of capabilities, thus maximizing the summarizer’s power.
In conclusion, the innovativeness of our text summarizer includes how it utilizes GPT-3 to make the text summarization accurate and reliable, as well as how it provides additional features that greatly enhance the summarization experience.
Ambitiousness - for the scale and complexity of solution
The ambitiousness of our text summarizer is precisely seen through how it achieves the aforementioned summarization objectives with features distinguishing it from normal text summarizers.
Firstly, the translation feature allows users to obtain summaries in languages different than that of the original text. Choosing a familiar and preferred language allows users to understand the text better and quicker, exactly aligning with the purpose of text summarization. Moreover, different from traditional translators which require users to open separate websites and therefore result in more data consumption, our text summarizer offers direct translation features for the summaries. Given that Ekatra is specifically designed for the Indian student population, Hindi translations are of course supported; in addition, the remarkable number of translatable language choices proves our text summarizer to be useful not only locally in India, but also globally.
Secondly, our text summarizer can work with directly inputted texts, or extract texts from web article links or uploaded documents. Educators and students do not need to worry about file conversions before uploading them for summarization - the file type will not serve as a constraint to the summaries.
Thirdly, our text summarizer allows users to customize the lengths of summaries by specifying a word limit, best satisfying users’ needs by allowing them to tailor the number of details needed in their summaries.
In summary, our text summarizer has a series of distinct features proving itself useful in both local and global contexts, respectively demonstrating the complexity and scale of the solution we developed.
Applicability – for adoption in the real-world
Universal access to quality education is a challenge to societies across the globe. SDG4 of the United Nations focuses on Quality Education, an objective whose importance has been widely recognized and understood, yet inequalities in education still prevail. With the not-yet-ceasing pandemic and the shift toward digitalized learning, gaps in education further enlarge due to unequal access to technologies.
Applying the text summarizer we developed to the Ekatra platform helps to further minimize the amount of data required for learning. When browsing through the large volume of texts, students can select and pinpoint sources they need based on the summaries instead of the whole texts, saving time and data, both being precious resources for all people, particularly Ekatra’s target audiences. Our summarizer and its added-on features provide an implementable solution that is applicable to Ekatra students or any students around the globe whose education experience is largely limited by internet access. It aims to minimize gaps in education by smoothing the impacts of uneven access to resources.
However, the utility it provides is not limited to the scope of Ekatra. With our text summarizer, researchers can reduce their perusing time during source searching processes; educators can extract key information easier; businesses can compile file archives efficiently. It even has the advantage of generating less-biased summaries than those done by humans.
Anyone with any objective in mind would find the text summarizer applicable and useful, with customized language and length. It surely provides unneglegible benefits to the real world.