Non-Linear Classification of Text Samples
AssigneeBazaarvoice, Inc.
InventorsDaniel Mahler, Eric D. Scott, Milos Curcic, Eric Allen
Life of this patent
Estimating the term…
Abstract
Non-linear classifiers and dimension reduction techniques may be applied to text classification. Non-linear classifiers such as random forest, Nyström/Fisher, and others, may be used to determine criteria usable to classify text into one of a plurality of categories. Dimension reduction techniques may also be used to reduce feature space size. Machine learning techniques may be used to develop criteria (e.g., trained models) that can be used to automatically classify text. Automatic classification rates may be improved and result in fewer numbers of text samples being unclassifiable or being incorrectly classified. User-generated content may be classified, in some embodiments.
Claims20
A method, comprising:
- a computer system processing each text sample in a training set, wherein each text sample in the training set corresponds to one of a plurality of classifications, and wherein processing each text sample includes:
- generating a respective set of features from that text sample;
- populating an entry in a data structure based on results of one or more dimension reduction operations performed on ones of the respective set of features for that text sample;
- the computer system using a non-linear classifier on entries in the data structure to establish criteria usable to classify an unknown text sample into one of the plurality of classifications.
A non-transitory computer-readable medium storing program instructions executable by a computing system to perform operations comprising:
- for each text sample within a training set for a plurality of different classifications, performing one or more dimension reduction operations on features generated from the text sample to populate a corresponding entry for that text sample in a data structure;
- and
- using a non-linear classifier to operate on entries in the data structure to establish criteria usable to classify an unknown text sample into one or more of the plurality of different classifications.
A method, comprising:
- a computer system classifying an unknown text sample into one of a plurality of categories, wherein the classifying comprises:
- generating a set of features from the unknown text sample;
- performing one or more dimension reduction operations on the generated set of features;
- and
- using a first set of criteria and data generated from the performing the one or more dimension reduction operations to determine a classification for the unknown text sample, wherein the first set of criteria was established by a training procedure that includes:
- performing one or more dimension reduction operations on features within text samples in a training set that includes text samples for each of the plurality of categories;
- populating a data structure having entries corresponding to text samples in the training set, wherein the populating is based on the one or more dimension reduction operations;
- using a non-linear classifier on the data structure to establish the first set of criteria.
Dependencies are inferred from the claim text.