/Text2Event

Primary LanguageJupyter Notebook

Text2Event: Event detection from domain-specific text data

A survey paper for overview: Wei, Xiang & Wang, Bang. (2019). A Survey of Event Extraction from Text. IEEE.

https://www.researchgate.net/publication/337638438_A_Survey_of_Event_Extraction_from_Text

WAMEX Dataset

https://www.dmp.wa.gov.au/WAMEX-Minerals-Exploration-1476.aspx

Files

1. Text classification using different machine learning algorithms

source/20news_classification.ipynb

alt_text Source: https://blogs.sas.com/content/subconsciousmusings/files/2017/04/machine-learning-cheet-sheet-2.png

2. source/extract_event_example.ipynb is a simple example file to extract events using pattern matching (trigger words)

  • Detect events using trigger words
  • Extract event details

3. source/graph_construction_from_text.ipynb does inormation extraction and graph construction

  • Extract entities using a domain dictionary source/VOCABULARY_TYPED.json (https://github.com/majiga/OzROCK labelled dataset can be used for supervised named entity recognition; Dictionary folder also contains the 6 lists of entities.)
  • Extract relations
  • Build a graph and visualise

alt_text

4. source/visualisation_WA.ipynb visualises the data/events.csv file on the WA map

  • Read data/events.csv file alt_text

  • Visualise the events on the WA map (data/map_WA.csv file can be useful for getting latitude and longitude data) alt_text

Visualisation examples

Some graph visualisation that were created from mineral exploration reports: https://sites.google.com/view/majiga/case-studies

Graph from text demo: https://nlp-tlp.org/text2kg/

NLP tools:

spacy.io

Outcomes

  1. List of events with their details
  2. Visualisations to explore or discover events
  3. Analysis, ...
  4. Check other required outcomes stated by the unit in https://handbooks.uwa.edu.au/unitdetails?code=CITS5553

alt text