Generate Your L(AI)brary Hackathon 2023

These workshops are all optional to attend, but highly recommended!

HCI Resources

• HCI x AI Considerations by Marc Astbury, Lead Designer @ Jenni.ai and Mentor for Generate Your L(AI)brary Hackathon 2023

Python Packages

Below is a list of popular python packages that can make building your prototype easier! These packages are specifically designed with Generative AI applications in mind, either by making it easy to create user interfaces or by making it easier to use Large Language Models.

• ChainLit - A package that provide a ChatGPT-like user interface in just a few lines of code.
• StreamLit - Similar package to ChainLit, but it has more customizable options such as slider, radio buttons, images, etc.
• LangChain - This package provides more structured and convenient ways to connect to popular models in OpenAI, HuggingFace, and others.
• LlamaIndex - Another package similar to LangChain but with less granular control. Both are great for summarization and Q&A tasks. See comparison of LangChan vs LLamaIndex.

APIs

• Alma API - Alma is our library system and you can retrieve the list of resources owned by SMU Libraries such as books, journals, databases, etc (we call them "holdings") for short. We will provide an API key on request
  • Check out this article for a brief overview of how it works and some use cases: “You could use the API!”: A Crash Course in Working with the Alma APIs using Postman
• Primo API - You can use this API to conduct a search of our library catalog. This essentially gets you results from our library default search
  • You can see a sample application here: Create an GPT-based Chatbot on Exlibris Knowledge Center. We will provide an API key on request
• OpenAlex API - Free and easy to use API, no API key needed. This gives you access to the largest source of metadata (Title, author, abstract) of Scholarly content. However, there is no full text. - High recommended if you need a source of Scholarly content
• CORE API - This API enables you to search all of the open access articles out there and access the full-text. Open access here means the article is freely available and not blocked by paywall. See CORE-GPT app. You need to register for an API key in advance.
• Scopus API - This API gives you articles results from Elsevier's Scopus database. Unlike the other APIs, this is restricted to only journal articles and only from prestigious journals. Scopus API also allows you to get citation counts and this is also the source behind some of the weights in University Ranking Systems by Times Higher Education and QS rankings. You need to register for API key and institutional token in advance. 
• Tutorial: Using APIs with Power Automate

Datasets

(Please note that below are just sample of anonymized data. If you'd like the full dataset to help with your hackathon project, please let us know!)

• EZproxy log - EZproxy is our proxy system that authenticates users prior to granting access to library e-resources such as articles, ebooks, and more. This log has been anonymized and captures the activity for the past month.
• Research Guides XML export - The dataset comprises the XML export of our publicly accessible research guide pages.
• Chat logs - Txt Export - This anonymized dataset comprises of library chat logs.

No-code AI bot creator

• Voiceflow - Voiceflow is a no-code tool collaborative platform to create AI assistants. Check out their Youtube channel for tutorials! The sandbox version is free with some limited options.

Not sure which problem to tackle in the Hackathon? See this article for some ideas for projects relating to SMU Libraries and research

Here's a few ideas!

1. Intelligent recommendation system where it will recommend a resource (e.g. workshops, article, etc) based on a profile e.g. modules taken, schools, etc.
2. Targeted EDM (for workshops or other things) based on user profile
3. AI chatbots used to man desk counters or chat services or after office hours. (See first library bot example or SMU Telegram bot example)
4. Q&A answer search engine that uses search to retrieve relevant answers which are then used by generator to answer the question (Also known as Retriever augmented generation) 
5. Determine whether an email from a library user is urgent
6. Summarise a list of articles or readings
7. Analyse sentiment of audio or video recordings
8. Anonymise data from surveys, focus groups, or interviews
9. Auto-tag the topics in an interview transcript in qualitative research
10. Auto-tag images based on what’s inside
11. Generate a report or summary on a given statistics (e.g. loan stats or workshop attendance)
12. Label and summarise library collection according to UN 17 SDG Goals, where applicable.
13. Turn book summary into colour, e.g. green for environmental related.
14. Done up citation for any given article / speech / presentation / journal / etc.
15. Catalogue / provide metadata by scanning through a book’s TOC
16. an AI character who narrates library services and answer questions
17. Research, track and recommend trending and useful library workshops offered for university students worldwide and map them to SMU Graduate Learning Output (GLO)
18. Create a ChatGPT plugin(see documentation, requires ChatGPT plus) that searches with SMU libraries' resources API such a Scopus, Web of Science
19. Automate a workflow to increase productivity via tools like Zapier, PowerAutomate and OpenAI API