-
-
Simply upload your file, choose a theme, click generate, and wait for your work to be done for you!
-
Downloads are quick and easy!
-
We run on all modern browsers and mobile devices!
What it does
EZPPT runs Natural Language Processing algorithms that are able to analyze a DOCX document and return a PowerPoint file with the most important points summarized into a neat PowerPoint template that is ready to use with only a few tweaks from the user.
Inspiration
Time is currency in our new technological age, and time should be put to good use. This is what EZPPT is all about, lost time working on creating weekly powerpoints for reports can be put for better use in deep thinking and solving problems. On average an employee spends 8 hours per month preparing powerpoints for weekly or monthly meetings and assuming a base wage of $15/hour, then we have around $1440/employee lost to PowerPoint building that could be used to a way better use. For a company with around a hundred employees, that is 6 figures lost in revenue. We all have passed through the tedious and repetitive work of creating a deck for reports written, and this is what inspired us to create EZPPT.
How we built it
The web app was built using Python Flask as the bone backend. This was built using 3 main APIS: Python-docx, which is used to be able to convert the docx into a list of paragraphs divided by headings to be able to analyze each paragraph individually and summarize it. The NLTK module is used to write the summarization algorithm that summarizes the paragraph depending on how relevant it is to the topic of the paragraph. This is done by finding the weight (importance) of each sentence which is done after the weight of each word is counted to how important it is to the meaning of the paragraph. Finally, the pptx-python module takes as input the summarized paragraphs and creates a PowerPoint slide with a user-selected theme, and allows the user to download it and edit it as he wishes.
Challenges we ran into
Our first goal was to take as input a PDF file rather than a DOCX file but then we realized that it is nearly impossible and we had to change our tech stack. This occurred since a PDF file does not have a proper style but all the text in it is positional (like a picture), therefore we needed to implement an OCR algorithm to capture the text, which would take too much time and memory for a hackathon.
Accomplishments that we're proud of
Writing and running the NLP algorithm is what we are proud of most since it took us the most time to be able to create a summarization algorithm with the perfect parameters to be able to summarize different documents about different topics.
What we learned
We learned that NLP algorithms are harder than they look at first glance especially if it needs to work with a whole language without a certain topic given as input. Also, reading from a file is harder and not documented as needed, which was weird with the booming of data science. Which has pushed us to work on python modules to solve this problem in the future.
What's next for EZ PowerPoint
EZ Powerpoint is a good business opportunity since it can attract hundreds of thousands of users per month, and integrating Adsense and ads, in general, can lead to a very fair revenue. EZ Powerpoint's next step is to optimize the summarizing algorithms so that a user does not even edit it after, integrate adding photos to the PowerPoint, and building a scalable website to launch to the public. In the future, it also could do the reverse, create a full report from a PowerPoint.
Log in or sign up for Devpost to join the conversation.