High-Level Project Summary
Smart Analyzer is an extremely intelligent AI powered application. It provides a feature that is capable of accurately summarizing research/scientific documents in the PDF format. The user can also obtain the summaries in a preferred number of sentences in a preferred language. 60 major languages are supported. It offers a feature that enables users to extract keywords and/or phrases from large documents containing highly complicated information. The voice assistant enables voice operation and can answer questions. Smart Analyzer fully addresses the challenge of building a solution to summarize NTRS documents and extract keywords. It additionally provides translation and the voice assistant.
Link to Final Project
Link to Project "Demo"
Detailed Project Description
How it Works
Smart Analyzer is a .NET Core UWP (Universal Windows Platform) application written in C# and XAML. All the AI services involved are provided by Microsoft azure. The document summarization feature can accurately summarize large PDF research documents. Summarization is done by an algorithm constructed with the Summarization feature/API of the Azure Language Service (a unified service for language by Azure). The Bing Spell check is involved in correcting spelling and grammatical errors in the summary that may exist. The Azure translator service translates the summary to a preferred language selected by the user. The Key-phrase extraction API of the Azure Language Service extracts keywords and important phrases. The NLP algorithm that operates the voice assistant communicates with Azure LUIS to determine the appropriate response. Text to speech and speech to text services are also employed by the voice assistant.
Importance of Research Data
Research is one of the most, if not the most, important aspect of scientific development that drives the modern world that we live in. The future is pulled into the present and immediately thrown into the past. As the future comes to us in this manner, it brings its problems with it, and we, are to solve them. This is where research, the process of fueling thought and building solutions comes into play. Regardless of the gender, nationality, race, religion, skin color, social status, country, region or any other form of division, the value of research applies to the entire world since everything has a future, and no one can escape it.
Scientists, researchers and thinkers worldwide need access to data. Above all, firstly, they must be able to understand information, and secondly, rapidly find exactly what they need, as they have no time to waste. How can this be achieved? This is the question.
How does Smart Analyzer answer this?
Simple answer. Smart Analyzer unleashes the true combined power of Artificial Intelligence and developments in Natural Language Processing. It simply enables these individuals to know exactly what they need to know about a particular collection of information/data (in other words, a research document). Also these valuable contributors will not be affected by language barriers. Therefore, Smart Analyzer unifies knowledge and reason under one platform.
Space Agency Data
In order to test Smart Analyzer, PDF research documents that were downloaded from the NTRS (NASA Technical Reports Server) were used. The documents were heavily involved in testing the summarization and key phrase/word analysis features of the application to ensure the accurate functioning of these two features. The documents contributed greatly to the identification of errors, vulnerabilities in the relevant algorithms and the factors that required improvement.
Hackathon Journey
This is my first time participating in a Space Apps challenge. It's a hackathon of a unique type in my opinion since challenges are extremely specific and the critical thinking ability of participants are assessed on a whole new level. I chose the challenge "Can AI preserve our science legacy" because the rapid spreading of understanding about scientific research is essential for the modern world and AI can enhance it more than any other technological factor. I would like to immensely thank NASA , the US government and the other partners for the creation of this golden opportunity.
In terms of building the solution, it was tough. Throughout the development process, I came across countless annoying issues including various errors, compatibility issues and mysterious failures by some of the algorithms that were absolutely haunting at the time. Spending more time on resolving the problems that were thrown at me and trying out different alternatives always worked in this journey. It was a quite rare type of a mind training I would say. It makes you realize again and again that patience is key and that answers for everything is reason, you just have to figure out the reasoning that is relevant to the specific case. That is the most important lesson of all, not the facts that were learnt. Now, I enjoy looking at the sweet result of my effort very much.
References
Coding support was obtained by Microsoft Documentations and Sample code.
Links:
https://learn.microsoft.com/en-us/azure/cognitive-services/translator/translator-text-apis?tabs=csharp#translate-text
https://learn.microsoft.com/en-us/azure/cognitive-services/language-service/summarization/quickstart?source=recommendations&tabs=document-summarization&pivots=programming-language-csharp
https://learn.microsoft.com/en-us/azure/cognitive-services/language-service/summarization/how-to/document-summarization
https://learn.microsoft.com/en-us/azure/cognitive-services/translator/translator-text-apis?tabs=csharp#translate-text
https://australiaeast.dev.cognitive.microsoft.com/docs/services/
Tags
#ArtificialIntelligence #NLP #.NET #UWP #C# #Azure