SimplyReading

Accomplishments that we're proud of

We are proud that SimplyReading targets the actual process of reading the entire source material as opposed to abbreviating it. Resources like SparkNotes or automated software deliver abridged versions, but they do so by sacrificing detail. With SimplyReading, we preserve content and information but convey it in a more understandable avenue.

Inspiration

During our brainstorming session, we discussed the topic of standardized testing, more specifically the SAT reading comprehension subsection that we all dread. We addressed the difficulty in improving reading comprehension and the frustration in reading texts over and over again only to have grasped a weak understanding of the information. We were inspired to create software that can tackle this all-too-common issue amongst students of all ages. We wanted to build a reading aid that wouldn’t sacrifice details and content by mere summary, but instead, it would make reading the existing information easier

What it does

SimplyReading takes any text and rewords it so that the information is easier to understand. After uploading a text file or typing in phrases, our web app will analyze each sentence down to the individual words. With natural language processing, we generated metrics to assess alternative phrases. Our program analyzes the generated versions, allowing users to simplify any text down to their individual needs from K-12.

How we built it

Our hack is split into two main components: our front-end web app using React JS and our backend using python. The bread and butter of our hack handled all textual analysis and regenerations in the backend. Given any sort of text, our program needed to measure each word’s complexity and pick proper replacements. Because there weren’t any existing libraries that could handle all this information, we engineered our own metric. With hours of research along with trial and error, we finally developed our own metric that combined word difficulty, frequency, and similarity. Through GCP, RapidAPIs, and Kaggle datasets, the metric outputted a quantifiable “weight.” In the end, use the weight to produce a new version of the original text based on the user’s preference of reading level.

Challenges we ran into

The most frustrating roadblock was the lack of relevant information and resources available to us that involves simplifying text complexity. We could not find any standardized or reliable method to quantify or modify text complexity. Any affordances we did have turned out to be weak analysis tools. In response, we devised our own solution to measure and reword text complexity using natural language processing.

What's next for SimplyReading

We foresee two steps of action for SimplyReading. First, we want SimplyReading to enter the hands of the user by developing our mobile app. We are always on the go, and we resort to quick entertainment like Twitter or Tik Tok. We are attracted to these platforms because they are easy and intuitive to use. By mobilizing our app, users can consume more information beyond social media. Second, SimplyReading can be used to reword source text to generate more complex sentence structures. As a digital classroom aid, this would help improve students’ comprehension of denser readings.

What we learned

One of our team members participated in his first hackathon, and he learned that pulling an all-nighter is not equivalent to more productivity. More importantly, though, both team members learned how to use the Google Cloud Platform’s Natural Language Processing API to extract a plethora of data points for any given set of sentences. As well as other various libraries and APIs, we struggled but figured out how to piece these together. It was an amazing experience in learning to use new APIs and technologies to build SimplyReading that is relevant to both ourselves and our communities.

Built With

• css
• gcp
• googlenlp
• javascript
• kaggle
• machine-learning
• python
• rapidapi
• react
• react-material
• twinword
• wordapi