alienDB-public

Code repository of the https://alienDB.org website.

Overview

alienDB.org ("AlienDB") is a data and ML-centric website about ET life, UAPs, SETI, METI and the human beings behind the search. The site is only a few weeks old (as of mid August, 2022) so it's incomplete, rapidly evolving, and fairly messy.

For now I think it will be best if I just include a few of the most useful snippets and perhaps the source of the homepage. Once the site and its data tables are more mature and stable I will see about posting the full, reproducible code.

Contents

I picked out the following code snippets as being the most interesting:

1. News data scrapers (moved from R to SerpAPI-based CURL commands)
2. Text Mining
3. Gantt-style Interative Timeline
4. Vertical timeline plot (which is static, for now)

Comments on Streamlit

This is the first site I've written using Streamlit, asides from a "Hello World" and reproducing some examples from their gallery. I've written webapps and portals using Python's Django framework and compared to that I would say that Streamlit ertainly caters more to data scientists, although Django provides a closer mapping to the underlying HTML, CSS & JS (and Model-View-Whatever) which is advantageous for writing and maintaining multiuser platforms / portals.

My choice was primarily based on what I believed to be syngery with my (real) job. Overall Streamlit has been easy to use and performs adequately.

Pro's of Using Streamlit

• It makes it easy to work with Python-based data
• The employees of the company are pretty good about replying to the questions they can at https://discuss.streamlit.cio
• No need to make a purchase
• Simplifies making a website while still allowing the flexibility of writing code

Con's of Using Streamlit

• Relative to Shiny, it's less flexible, has far fewer examples, and a smaller community - but I expect this to be a temporary situation as Streamlit expands and its users gain more experience (and publish more of their code)
• Although it claims to be fast even with Big Data, on the small scale it can be slower than expected
• Each page is an app which must be reloaded and run on every page load
• Streamlit uses your website as free advertising ("Made with Streamlit") and even if you hide this with CSS it will still show up while things are loading and on errors

Conclusion

I think Streamlit is a great tool and I expect it to keep getting better. If this site were not in part a Streamlit-learning excercise I think a better approach might be to write non-data-centric components of the website using native web technologies and limit Streamlit to serving just 1 or 2 pages of data products. Bear in mind that I'm still a Streamlit noob, so my perceptions may be shaped by my limited experience and expertise.