Unsupervised neural attention model for aspect extraction
Methods, systems, and computer-readable storage media for receiving a vocabulary, the vocabulary including text data that is provided as at least a portion of raw data, the raw data being provided in a computer-readable file, associating each word in the vocabulary with a feature vector, providing a sentence embedding for each sentence of the vocabulary based on a plurality of feature vectors to provide a plurality of sentence embeddings, providing a reconstructed sentence embedding for each sentence embedding based on a weighted parameter matrix to provide a plurality of reconstructed sentence embeddings, and training the unsupervised neural attention model based on the sentence embeddings and the reconstructed sentence embeddings to provide a trained neural attention model, the trained neural attention model being used to automatically determine aspects from the vocabulary.
1. A computer-implemented method for aspect extraction from raw data using an unsupervised neural attention model, the method being executed by one or more processors and comprising:
receiving, by the one or more processors, a vocabulary, the vocabulary comprising text data that is provided as at least a portion of raw data, the raw data being provided in a computer-readable file;
associating, by the one or more processors, each word in the vocabulary with a feature vector;
providing, by the one or more processors, a sentence embedding for each sentence of the vocabulary based on a plurality of feature vectors to provide a plurality of sentence embeddings;
providing, by the one or more processors, a reconstructed sentence embedding for each sentence embedding based on a weighted parameter matrix to provide a plurality of reconstructed sentence embeddings; and
training, by the one or more processors, the unsupervised neural attention model based on the sentence embeddings and the reconstructed sentence embeddings to provide a trained neural attention model, the trained neural attention model being used to automatically determine aspects from the vocabulary, training comprising obtaining an objective function based on an unregularized objective, and a regularization term.
2. The method of claim 1 , wherein each sentence embedding is provided by applying a respective weight to each feature vector, the respective weight indicating a probability that a respective word captures an aspect of the respective sentence, the weight being determined by an attention model.
3. The method of claim 1 , wherein the unregularized function is obtained based on the sentence embeddings and the reconstructed sentence embeddings.
4. The method of claim 1 , wherein each reconstructed sentence embedding is provided by dimension reduction and reconstruction of the respective sentence embedding.
5. The method of claim 1 , wherein each feature vector is provided as a word embedding.
6. The method of claim 1 , wherein the raw data comprises review data.
7. A non-transitory computer-readable storage medium coupled to one or more processors and having instructions stored thereon which, when executed by the one or more processors, cause the one or more processors to perform operations for aspect extraction from raw data using an unsupervised neural attention model, the operations comprising:
receiving a vocabulary, the vocabulary comprising text data that is provided as at least a portion of raw data, the raw data being provided in a computer-readable file;
associating each word in the vocabulary with a feature vector;
providing a sentence embedding for each sentence of the vocabulary based on a plurality of feature vectors to provide a plurality of sentence embeddings;
providing a reconstructed sentence embedding for each sentence embedding based on a weighted parameter matrix to provide a plurality of reconstructed sentence embeddings; and
training the unsupervised neural attention model based on the sentence embeddings and the reconstructed sentence embeddings to provide a trained neural attention model, the trained neural attention model being used to automatically determine aspects from the vocabulary, training comprising obtaining an objective function based on an unregularized objective, and a regularization term.
8. The computer-readable storage medium of claim 7 , wherein each sentence embedding is provided by applying a respective weight to each feature vector, the respective weight indicating a probability that a respective word captures an aspect of the respective sentence, the weight being determined by an attention model.
9. The computer-readable storage medium of claim 7 , wherein the unregularized function is obtained based on the sentence embeddings and the reconstructed sentence embeddings.
10. The computer-readable storage medium of claim 7 , wherein each reconstructed sentence embedding is provided by dimension reduction and reconstruction of the respective sentence embedding.
11. The computer-readable storage medium of claim 7 , wherein each feature vector is provided as a word embedding.
12. The computer-readable storage medium of claim 7 , wherein the raw data comprises review data.
13. A system, comprising:
a computing device; and
a computer-readable storage device coupled to the computing device and having instructions stored thereon which, when executed by the computing device, cause the computing device to perform operations for aspect extraction from raw data using an unsupervised neural attention model, the operations comprising:
receiving a vocabulary, the vocabulary comprising text data that is provided as at least a portion of raw data, the raw data being provided in a computer-readable file;
associating each word in the vocabulary with a feature vector;
providing a sentence embedding for each sentence of the vocabulary based on a plurality of feature vectors to provide a plurality of sentence embeddings;
providing a reconstructed sentence embedding for each sentence embedding based on a weighted parameter matrix to provide a plurality of reconstructed sentence embeddings; and
training the unsupervised neural attention model based on the sentence embeddings and the reconstructed sentence embeddings to provide a trained neural attention model, the trained neural attention model being used to automatically determine aspects from the vocabulary, training comprising obtaining an objective function based on an unregularized objective, and a regularization term.
14. The system of claim 13 , wherein each sentence embedding is provided by applying a respective weight to each feature vector, the respective weight indicating a probability that a respective word captures an aspect of the respective sentence, the weight being determined by an attention model.
15. The system of claim 13 , wherein the unregularized function is obtained based on the sentence embeddings and the reconstructed sentence embeddings.
16. The system of claim 13 , wherein each reconstructed sentence embedding is provided by dimension reduction and reconstruction of the respective sentence embedding.
17. The system of claim 13 , wherein each feature vector is provided as a word embedding.