I tried to do something similar as one of my final year project with something similar ( keyword and category monitoring ) with a future H-index prediction ranking model, which I think is a more unique approach. The issue with arxiv monitoring is that the only way to obtain the latest submission is through their RSS feed ( correct me if i am wrong ) which luckily I have a running scraper service handling for myself. However the abstract and author from RSS feed isn’t enough, you need to parse PDF for meta info and conference submission as well. As a researcher I think we usually are more interested in new approach or ideas which definitely bring quite a lot of attention until it just pops up around social media ( twitter usually). If you simply want keyword notification service google scholar and semantic scholar already did it quite well. Semantic scholar also have their own API which is slightly better than Arxiv source