Skip to content

Latest commit

 

History

History
56 lines (38 loc) · 1.81 KB

README.md

File metadata and controls

56 lines (38 loc) · 1.81 KB
Raw

WikiQuery

Introduction

This is the database course project, which is about storing information in wikidata in SQL schemas.

It requires smart design on database schema, and possible optimization by indexing.

By implementing required queries and possible natural language queries, the database is essientially served as a knowledge graph.

To demonstrate the work, a web GUI will be designed and implemented.

Setup

For SQL table construction, run kg_simplified.sql in MySQL.

For data import, run python preprocess.py. Note about some global config in it.

For web display, we use flask for backend logic, react-js and materialize for front display:

Run the web app:

cd web/
python run.py

Remember to set up the sql connection in web/app/src/query_api.py

Architecture

.
├── data
│   └── wikidata-latest-all.json  	# omited due to large in size
├── web								# backend logic + GUI display 
├── util							# possible util functions 
├── resource						# other files
└── README.md

TODO

  • Loading event.

  • NLQ: several small queries

  • Qualifier in query

  • Website build for listen

  • Tree: Some "instance of" not included! Type "Fantasy" (依然范特西) Many entities fail to find any tree entity...

  • Co-occurred: Sort in eid order!

  • Statements: Sort them in Property order!

  • NLQ: Find_claim_value(eid, pid) is needed!

  • Add "#properties" in table "entity" for choosing from name.

Note

Wikidata can be retrieved here, which is approx 4.5G in compressed form, 80G when extracted.