Inspiration
Our inspiration for InstantResearch.AI came from the frustration of working with traditional academic archive searches. We noticed the subpar results, biased ranking, and outdated keyword methods that hinder users from finding relevant research papers. We wanted to transform this experience, making it efficient and effective for everyone.
What it does
InstantResearch.AI reorganizes academic archive search results using Cohere's semantic embeddings to make them more relevant and user-friendly. It inputs the topic and research, generates refined search criteria, queries the site, and reorders the results using Cohere rerank. To save users time and effort, it also provides high-level summaries of the selected articles.
How we built it
We built InstantResearch.AI using advanced AI and machine learning techniques, particularly Cohere's semantic embeddings. The system's query and ranking process are redefined to focus on semantic relevance rather than traditional keyword-based search methods.
Challenges we ran into
We faced challenges in fine-tuning the semantic embeddings to accurately understand and rank academic content. Moreover, summarizing complex academic articles into high-level, understandable summaries was another significant hurdle.
Accomplishments that we're proud of
Successfully integrating Cohere's semantic embeddings and creating a more effective and efficient search system for academic archives is a significant accomplishment. Also integrating Claude to summarize large-form articles to help a user understand the article they might read better. We're proud that InstantResearch.AI can save users considerable time and effort, making research more accessible.
What we learned
We learned a great deal about semantic embeddings and their potential applications in improving search algorithms. How to integrate Cohere’s re-rank function and the logic it uses. Using prompts to preprocess the user's inputs to get better data to process with algorithms. We also understood the importance of user-friendly interfaces in the context of academic research.
What's next for InstantResearch.AI
Moving forward, we plan to extend InstantResearch.AI's functionality to other domains like identifying similar patents during filing, helping lawyers find relevant legal cases, and internal knowledge base search. We also aim to continually refine our algorithm to provide even more accurate and helpful search results.
Log in or sign up for Devpost to join the conversation.