Keyword Extraction from Short Texts with T5

Our vlT5 model is a keyword generation model based on encoder-decoder architecture using Transformer blocks presented by Google (https://huggingface.co/t5-base). The vlT5 was trained on scientific articles corpus to predict a given set of keyphrases based on the concatenation of the article’s abstract and title. It generates precise, yet not always complete keyphrases that describe the content of the article based only on the abstract.

Keywords generated with vlT5-base-keywords: encoder-decoder architecture, keyword generation

Results on demo model (different generation method, one model per language):

Our vlT5 model is a keyword generation model based on encoder-decoder architecture using Transformer blocks presented by Google (https://huggingface.co/t5-base). The vlT5 was trained on scientific articles corpus to predict a given set of keyphrases based on the concatenation of the article’s abstract and title. It generates precise, yet not always complete keyphrases that describe the content of the article based only on the abstract.

Keywords generated with vlT5-base-keywords: encoder-decoder architecture, vlT5, keyword generation, scientific articles corpus

vlT5

The biggest advantage is the transferability of the vlT5 model, as it works well on all domains and types of text. The downside is that the text length and the number of keywords are similar to the training data: the text piece of an abstract length generates approximately 3 to 5 keywords. It works both extractive and abstractively. Longer pieces of text must be split into smaller chunks, and then propagated to the model.

Overview

Language model: t5-base
Language: pl, en (but works relatively well with others)
Training data: POSMAC
Online Demo: Visit our online demo for better results https://nlp-demo-1.voicelab.ai/
Paper: Keyword Extraction from Short Texts with a Text-To-Text Transfer Transformer, ACIIDS 2022

Corpus

The model was trained on a POSMAC corpus. Polish Open Science Metadata Corpus (POSMAC) is a collection of 216,214 abstracts of scientific publications compiled in the CURLICAT project.

Domains	Documents	With keywords
Engineering and technical sciences	58 974	57 165
Social sciences	58 166	41 799
Agricultural sciences	29 811	15 492
Humanities	22 755	11 497
Exact and natural sciences	13 579	9 185
Humanities, Social sciences	12 809	7 063
Medical and health sciences	6 030	3 913
Medical and health sciences, Social sciences	828	571
Humanities, Medical and health sciences, Social sciences	601	455
Engineering and technical sciences, Humanities	312	312

Tokenizer

As in the original plT5 implementation, the training dataset was tokenized into subwords using a sentencepiece unigram model with vocabulary size of 50k tokens.

Usage

from transformers import T5Tokenizer, T5ForConditionalGeneration

model = T5ForConditionalGeneration.from_pretrained("Voicelab/vlt5-base-keywords")
tokenizer = T5Tokenizer.from_pretrained("Voicelab/vlt5-base-keywords")

task_prefix = "Keywords: "
inputs = [
    "Christina Katrakis, who spoke to the BBC from Vorokhta in western Ukraine, relays the account of one family, who say Russian soldiers shot at their vehicles while they were leaving their village near Chernobyl in northern Ukraine. She says the cars had white flags and signs saying they were carrying children.",
    "Decays the learning rate of each parameter group by gamma every step_size epochs. Notice that such decay can happen simultaneously with other changes to the learning rate from outside this scheduler. When last_epoch=-1, sets initial lr as lr.",
    "Hello, I'd like to order a pizza with salami topping.",
]

for sample in inputs:
    input_sequences = [task_prefix + sample]
    input_ids = tokenizer(
        input_sequences, return_tensors="pt", truncation=True
    ).input_ids
    output = model.generate(input_ids, no_repeat_ngram_size=3, num_beams=4)
    predicted = tokenizer.decode(output[0], skip_special_tokens=True)
    print(sample, "\n --->", predicted)

Inference

Our results showed that the best generation results were achieved with no_repeat_ngram_size=3, num_beams=4

Results

Method	Rank	Micro			Macro
		P	R	F1	P	R	F1
extremeText	1	0.175	0.038	0.063	0.007	0.004	0.005
	3	0.117	0.077	0.093	0.011	0.011	0.011
	5	0.090	0.099	0.094	0.013	0.016	0.015
	10	0.060	0.131	0.082	0.015	0.025	0.019
vlT5kw	1	0.345	0.076	0.124	0.054	0.047	0.050
	3	0.328	0.212	0.257	0.133	0.127	0.129
	5	0.318	0.237	0.271	0.143	0.140	0.141
KeyBERT	1	0.030	0.007	0.011	0.004	0.003	0.003
	3	0.015	0.010	0.012	0.006	0.004	0.005
	5	0.011	0.012	0.011	0.006	0.005	0.005
TermoPL	1	0.118	0.026	0.043	0.004	0.003	0.003
	3	0.070	0.046	0.056	0.006	0.005	0.006
	5	0.051	0.056	0.053	0.007	0.007	0.007
	all	0.025	0.339	0.047	0.017	0.030	0.022
extremeText	1	0.210	0.077	0.112	0.037	0.017	0.023
	3	0.139	0.152	0.145	0.045	0.042	0.043
	5	0.107	0.196	0.139	0.049	0.063	0.055
	10	0.072	0.262	0.112	0.041	0.098	0.058
vlT5kw	1	0.377	0.138	0.202	0.119	0.071	0.089
	3	0.361	0.301	0.328	0.185	0.147	0.164
	5	0.357	0.316	0.335	0.188	0.153	0.169
KeyBERT	1	0.018	0.007	0.010	0.003	0.001	0.001
	3	0.009	0.010	0.009	0.004	0.001	0.002
	5	0.007	0.012	0.009	0.004	0.001	0.002
TermoPL	1	0.076	0.028	0.041	0.002	0.001	0.001
	3	0.046	0.051	0.048	0.003	0.001	0.002
	5	0.033	0.061	0.043	0.003	0.001	0.002
	all	0.021	0.457	0.040	0.004	0.008	0.005

License

CC BY 4.0

Citation

If you use this model, please cite the following paper: Pęzik, P., Mikołajczyk, A., Wawrzyński, A., Żarnecki, F., Nitoń, B., Ogrodniczuk, M. (2023). Transferable Keyword Extraction and Generation with Text-to-Text Language Models. In: Mikyška, J., de Mulatier, C., Paszynski, M., Krzhizhanovskaya, V.V., Dongarra, J.J., Sloot, P.M. (eds) Computational Science – ICCS 2023. ICCS 2023. Lecture Notes in Computer Science, vol 14074. Springer, Cham. https://doi.org/10.1007/978-3-031-36021-3_42

Piotr Pęzik, Agnieszka Mikołajczyk-Bareła, Adam Wawrzyński, Bartłomiej Nitoń, Maciej Ogrodniczuk, Keyword Extraction from Short Texts with a Text-To-Text Transfer Transformer, ACIIDS 2022

Authors

The model was trained by NLP Research Team at Voicelab.ai.

You can contact us here.

Voicelab
/

vlt5-base-keywords