andersonsacramento/fed repository overview

⁠FED

FED - Frame Event Detection

⁠DESCRIPTION

FED is a closed domain event detector system for sentences in the Portuguese language. It detect events from sentences, i.e., event trigger identification and classification. The event types are based on the typology of the FrameNet project (BAKER; FILLMORE; LOWE, 1998). The models were trained on an enriched TimeBankPT (COSTA; BRANCO,2012) corpus.

Currently, in this Colab, 5 different trained models are available to execution: 0, 5, 25, 50, and 100 which respectively correspond to: 214, 137, 31, 13, and 5 event types.

The system outputs the event detections in the following Json format:

[
    {
        "text": "aumentou",
        "start": 12,
        "end": 20,
        "event_type": "Cause_change_of_position_on_a_scale"
    },
    {
        "text": "disse",
        "start": 58,
        "end": 63,
        "event_type": "Statement"
    }
] 

The andersonsacramento/fed Docker image reads in an input directory of text files and writes Json document files to an output directory or it reads a sentence and prints the output on the standard output in Json format.

The text files in the input directory are expected to have the format:

* all text files end with the extension .txt
* sentences are separated by newlines

In order to read and write files on the Docker host, the container needs to be run with appropriate Docker volume⁠ flags (e.g. -v/--volume/--mount).

If the working directory on the Docker host contains a sub-directory named input-files/, then the container could be run using:

$ docker run -it --rm  -v  `pwd`:/mnt andersonsacramento/fed --dir /mnt/input-files/  /mnt/output-files/

In order to run the container on a given sentence, it can be invoked using:

$ docker run -it  --rm andersonsacramento/fed --sentence 'A Petrobras aumentou o preço da gasolina para 2,30 reais, disse o presidente.'

⁠How to cite this work

Peer-reviewed accepted paper:

10th Brazilian Conference on Intelligent Systems (BRACIS)

• Sacramento A. ; Souza M. . Joint Event Extraction with Contextualized Word Embeddings for the Portuguese Language.