This tool has the aim to let you classify citation intents. The classification part is carried out by an Ensemble Model, which is a combination of three binary classifiers and a meta classifier built on top of them.The meta classifier carries out the voting process and returns the final classification result. Furthermore, a threshold of 90% confidence has been defined to filter out the results on which the model is not confident enough.
The baseline model surpass the current SOTA Macro-F1 score for the citation intent classification task with SciCite dataset.
This tool gives you the possibility to classify any number of input sentences given in input in the form of a list of tuples. Clearly, higher the number of sentences, higher the time needed to classify all of them and to get back results. The tool also gives you the possibility to download the results in JSON format.
Finally, you have the possibility to select one of three possible working modes:
To use this tool, you have to go to the Classifier page and select the needed mode through the 3 buttons on top. Then, you have to insert your sentences in the form of a list of tuples, or as a JSON file. It is extremely important to follow the format of the input, otherwise the tool will not work.
Once you have added all the sentences you want to classify, simply click on the Classify button (or on the Classify JSON button if you uploaded a JSON file). The tool will carry out the classification process and return you the results. These results will be printed out on screen in an easy to understand format.
Additionally, at the end of the classification process, a Download JSON button will appear at the bottom of the page, by clicking on it the browser will download a JSON file containing the results. It is important to note that the time needed to classify all the sentences depends on the number of sentences you have inserted.