| import gradio as gr | |
| import nltk | |
| nltk.download('omw-1.4') | |
| from qanom.nominalization_detector import NominalizationDetector | |
| detector = NominalizationDetector() | |
| title = "Nominalization Detection Demo" | |
| description = f"""This is a demo of QANom's nominalization detection algorithm, | |
| comprised of candidate nominalization extraction followed by a contextualized binary classification model.""" | |
| links = """<p style='text-align: center'> | |
| <a href='https://github.com/kleinay/QANom' target='_blank'>QANom repo</a> | | |
| <a href='https://huggingface.co/kleinay/nominalization-candidate-classifier' target='_blank'>Model Repo at Huggingface Hub</a> | | |
| <a href='https://www.aclweb.org/anthology/2020.coling-main.274/' target='_blank'>QANom Paper</a> | |
| </p>""" | |
| examples = [["The doctor was interested in Luke 's treatment .", True, 0.5], | |
| ["The description of the horse 's jump provided a surprise to the owner and a show of the skill of the trainer .", True, 0.5], | |
| ["The construction of the officer 's building finished right after the beginning of the destruction of the previous construction .", True, 0.75]] | |
| def call(sentence: str, return_all_candidates: bool, threshold: float): | |
| ret = detector([sentence], return_all_candidates, True, threshold)[0] | |
| if return_all_candidates: | |
| positives = [d["predicate_idx"] for d in ret if d['predicate_detector_prediction']] | |
| negatives = [d["predicate_idx"] for d in ret if not d['predicate_detector_prediction']] | |
| else: | |
| positives = [d["predicate_idx"] for d in ret] | |
| negatives = [] | |
| def color(idx): | |
| if idx in positives: return "lightgreen" | |
| if idx in negatives: return "pink" | |
| idx2verb = {d["predicate_idx"] : d["verb_form"] for d in ret} | |
| idx2prob = {d["predicate_idx"] : d["predicate_detector_probability"] for d in ret} | |
| def word_span(word, idx): | |
| tooltip = f'title=" probability={idx2prob[idx]:.2}
verb={idx2verb[idx]}"' if idx in idx2verb else '' | |
| return f'<span {tooltip} style="background-color: {color(idx)}">{word}</span>' | |
| html = '<span>' + ' '.join(word_span(word, idx) for idx, word in enumerate(sentence.split(" "))) + '</span>' | |
| return html, ret | |
| iface = gr.Interface(call, | |
| inputs=[gr.inputs.Textbox(label="Sentence", lines=3), | |
| gr.inputs.Checkbox(default=True, label="Return all candidates?"), | |
| gr.inputs.Slider(minimum=0., maximum=1., step=0.01, default=0.5, label="Threshold")], | |
| outputs=[gr.outputs.HTML(label="Detected Nominalizations"), | |
| gr.outputs.JSON(label="Raw Model Output")], | |
| title=title, | |
| description=description, | |
| article=links, | |
| examples=examples ) | |
| iface.launch() |