Aspect extraction from product reviews with Tensorflow
This repo has multiple sequential models for aspect extraction from product reviews.
Citation
If the code is useful in your research, please cite the following paper:
Poria, S., Cambria, E. and Gelbukh, A., 2016. Aspect extraction for opinion mining with a deep convolutional neural network. Knowledge-Based Systems, 108, pp.42-49.
Task
Given a sentence, the task is to extract aspects. Here is an example
I like the battery life of this phone"
Converting this sentence to IOB would look like this -
I O
like O
the O
battery B-A
life I-A
of O
this O
phone O
Model
Similar to Ma and Hovy.
- concatenate final states of a bi-lstm on character embeddings to get a character-based representation of each word
- concatenate this representation to a standard word vector representation (GloVe here)
- run a bi-lstm on each sentence to extract contextual representation of each word
- decode with a linear chain CRF
Similar to [Collobert et al.] (http://ronan.collobert.com/pub/matos/2011_nlp_jmlr.pdf)
- form a window around the word to tag
- apply MLP on that window
- obtain logits
- apply viterbi (CRF) for sequence tagging
Similar to Poria et al.
- form a window around the word to tag
- apply CNN on that window
- apply maxpool on that window (Caution: different from global maxpool)
- obtain logits
- apply CRF for sequence tagging
XML to IOB
python xmlToIOB.py
Details
Download Glove embeddings (GloVe: http://nlp.stanford.edu/data/glove.840B.300d.zip )
- [DO NOT MISS THIS STEP] Build vocab from the data and extract trimmed glove vectors according to the config in
model/config.py.
python build_data.py
- Train the model with
python train.py
- Evaluate and interact with the model with
python evaluate.py
Data iterators and utils are in model/data_utils.py and the model with training/test procedures is in model/aspect_model.py
Training Data
The training data must be in the following format (identical to the CoNLL2003 dataset).
A default test file is provided to help you getting started.
The O
duck B-A
confit I-A
is O
always O
amazing O
and O
the O
foie B-A
gras I-A
terrine I-A
with I-A
figs I-A
was O
out O
of O
this O
world O
The O
wine B-A
list I-A
is O
interesting O
and O
has O
many O
good O
values O
Once you have produced your data files, change the parameters in config.py like
# dataset
filename_train = "data/ABSA16_Restaurants_Train_SB1_v2_mod.iob"
filename_dev = "data/EN_REST_SB1_TEST_2016_mod.iob"
filename_test = "data/EN_REST_SB1_TEST_2016_mod.iob"
Result
Chunk based evaluation
Laptop 2014 -> F1 - 79.93
Restaurant 2014 -> F1 - 84.22
Partial matching based evaluation
Laptop 2014 -> F1 - 86.84
Restaurant 2014 -> F1 - 88.42
License
This project is licensed under the terms of the apache 2.0 license (as Tensorflow and derivatives). If used for research, citation would be appreciated.