The unofficial Enron email corpus search engine.

1. Install Python 3.5 or above
2. Download and save the sample and full enron email corpus as enron and enron_mail_20150507, respectively.
3. Create a virtual environment and activate it before installing MACK.

Build and install the package.

python setup.py build && python setup.py install

Run MACK. Make sure to run MACK in the same directory where the email corpus is located.

mack --search app

• MOAR UNIT TESTS. The unit test coverage is very low, and I would want to get that at 100%.
• The MACK search engine does not score the documents retrieved from a query. I found a ton of choices while I was researching about search engines, and it would be a cool weekend project to add scoring functionality. Some example choices: BM25, PageRank, tf-idf.
• Reduce index build time. The build time takes about 12 seconds for the sample data set which is about 12mb. I briefly used Python's cProfiler and guppy to profile runtime and memory, but I felt it wasn't enough time to really find some optimizations. I could also take a step back and look at the architecture from a high-level and judge if it is really optimal.
• Support multi-term querying. You can provide multiple terms to MACK, but it will only use the first term to query the inverted index. When I was started the challenge, I wanted to support phrase querying but quickly realized it was extremely ambitious to support that in two days.

• https://nlp.stanford.edu/IR-book/html/htmledition/positional-indexes-1.html
• https://www.youtube.com/watch?v=cY7pE7vX6MU&t=1106s
• https://artem.krylysov.com/blog/2020/07/28/lets-build-a-full-text-search-engine/

• Q: What does MACK stand for? A: Absolutely, nothing.