For the Freelance

Inspiration

It is difficult for people to find right healthcare resources when they need them: information about the services from providers might be stored in different formats and spread out in different physical places. A virtual engine, which has a friendly human-machine interface and can reach to target information fast, is indeed needed.

What it does

It provides a comprehensive solution to build such a virtual engine:

  1. Speaking is more natural than reading and writing: a vocal chat-bot like Siri or Alexa will act as the interface to interact with people.
  2. The information needed is inconsistent and heterogeneous. The efficient and economical way to reach it precisely and fast is a theoretical robust and efficient search engine -- it can utilize the existing information systems rather than replace them, and can glue these different systems into a universal one.

How I built it

  1. The vocal chat-bot: adapt existing open-source projects, for example, Mycroft/Kalliope for Linux and Android, Stephanie is for Windows and Mac OSX.
  2. The search engine: first indexing all the available information using appropriate keywords, then build direct acyclic graphs (DAGs) using the keywords as internal nodes and target information as leaf nodes, at last indexing all the non-leaf nodes and connect with the vocal chat-bot.

Challenges I ran into

The unstable wireless network during the hackathon (kidding :)). The major challenges:

  1. Not familiar with the healthcare data.
  2. Did not even install Microsoft Office on my machine.

Accomplishments that I'm proud of

The idea about the DAG based search engine.

What I learned

The healthcare related data.

What's next for EasyAccess

Implementation:

  1. Periodically crawling related information, extract keywords, and do indexing. The raw data will be stored in HDFS, the indexing data (as DAGs) will be stored in a graphical database e.g. Neo4J, the processing will use NumPy/Pandas, the workflow management will use Apache Airflow.
  2. Based on the DAGs, a searching backend will be built using Flask micro framework in Python. All web services will be RESTful APIs.
  3. The first edition of UI will be a universal web portal, using Kendo UI for Angular or jQuery. They must work perfectly when without vocal chat-bot.
  4. Vocal chat-bot will be incorporated accordingly.

Built With

Share this project:
×

Updates