Automated email classification in an information management system
An improved information management system is described herein that can use artificial intelligence to classify data files (e.g., emails, documents, audio files, video files, etc.) and/or surface in a user interface the classification assigned to the data files. For example, a lightweight training or a heavyweight training can be employed to train classifiers to classify the data files. Use of artificial intelligence to classify data files may reduce the amount of computing resources that the information management system allocates to the data file review process because an auditor may be able to quickly identify those data files that meet the desired criteria using the classification and only request access to those data files. Thus, the information management system may be able to allocate more computing resources to normal or routine tasks, ensuring that such tasks are completed and/or completed in an appropriate amount of time.
1. A networked information management system for classifying data files, the networked information management system comprising:
a secondary storage device configured to store a plurality of data files; and
a classification system having one or more hardware processors, wherein the classification system is configured with computer-executable instructions that, when executed, cause the classification system to:
obtain a first subset of the plurality of data files;
train a multi-class classifier using the first subset of the plurality of data files;
obtain a second subset of the plurality of data files;
obtain content index data corresponding to the second subset of the plurality of data files;
for each data file in the second subset of the plurality of data files, tag the respective data file using at least a portion of the content index data; and
classify the tagged data files in the second subset of the plurality of data files using the multi-class classifier.
2. The networked information management system of claim 1 , wherein the computer-executable instructions, when executed, further cause the classification system to:
cause a secondary storage computing device to restore the first subset of the plurality of data files from the secondary storage device; and
train the multi-class classifier using the restored first subset of the plurality of data files.
3. The networked information management system of claim 1 , wherein the trained multi-class classifier is trained to assign one of a plurality of pre-defined classifications to a data file provided to the trained multi-class classifier as an input.
4. The networked information management system of claim 1 , wherein the plurality of pre-defined classifications comprises at least one of human resources, finance, legal, or personal.
5. The networked information management system of claim 1 , wherein the content index data comprises one or more keywords extracted from one or more of the first subset of the plurality of data files.
6. The networked information management system of claim 5 , wherein the computer-executable instructions, when executed, further cause the classification system to, for each data file in the second subset of the plurality of data files, tag the respective data file using at least one keyword in the one or more keywords extracted from the respective data file.
7. The networked information management system of claim 1 , wherein the computer-executable instructions, when executed, further cause the classification system to train the multi-class classifier using a lightweight approach.
8. The networked information management system of claim 7 , wherein each data file in the first subset of the plurality of data files is associated with a first tag and an actual classification, and wherein the computer-executable instructions, when executed, further cause the classification system to train the multi-class classifier using the first tags as features and the actual classifications as labels.
9. The networked information management system of claim 7 , wherein the computer-executable instructions, when executed, further cause the classification system to train the multi-class classifier using a k-nearest neighbor algorithm.
10. The networked information management system of claim 1 , wherein the computer-executable instructions, when executed, further cause the classification system to train the multi-class classifier using a heavyweight approach.
11. The networked information management system of claim 10 , wherein the computer-executable instructions, when executed, further cause the classification system to:
preprocess each data file in the first subset of the plurality of data files to extract a plurality of entities; and
train the multi-class classifier using the extracted plurality of entities as features.
12. The networked information management system of claim 11 , wherein the computer-executable instructions, when executed, further cause the classification system to:
for each data file in the first subset of the plurality of data files,
tokenize the respective data file;
filter the respective tokenized data file; and
lemmatize the respective filtered tokenized data file.
13. The networked information management system of claim 1 , wherein the computer-executable instructions, when executed, further cause the classification system to:
cause a secondary storage computing device to restore the second subset of the plurality of data files from the secondary storage device; and
classify the tagged data files in the restored second subset of the plurality of data files using the multi-class classifier.
14. The networked information management system of claim 1 , wherein the computer-executable instructions, when executed, further cause the classification system to train the multi-class classifier prior to reception of a request to classify the second subset of the plurality of data files.
15. A computer-implemented method for classifying data files, the computer-implemented method comprising:
obtaining a first subset of a plurality of data files;
training a multi-class classifier using the first subset of the plurality of data files;
obtaining a second subset of the plurality of data files;
obtaining content index data corresponding to the second subset of the plurality of data files;
for each data file in the second subset of the plurality of data files, tagging the respective data file using at least a portion of the content index data; and
classifying the tagged data files in the second subset of the plurality of data files using the multi-class classifier.
16. The computer-implemented method of claim 15 , wherein training a multi-class classifier further comprises:
causing a secondary storage computing device to restore the first subset of the plurality of data files from a secondary storage device; and
training the multi-class classifier using the restored first subset of the plurality of data files.
17. The computer-implemented method of claim 15 , wherein the trained multi-class classifier is trained to assign one of a plurality of pre-defined classifications to a data file provided to the trained multi-class classifier as an input.
18. The computer-implemented method of claim 15 , wherein the content index data comprises one or more keywords extracted from one or more of the first subset of the plurality of data files.
19. The computer-implemented method of claim 18 , wherein tagging the respective data file further comprises, for each data file in the second subset of the plurality of data files, tagging the respective data file using at least one keyword in the one or more keywords extracted from the respective data file.
20. A non-transitory computer readable medium storing computer-executable instructions that, when executed by a classification system, cause the classification system to:
obtain a first subset of a plurality of data files;
train a multi-class classifier using the first subset of the plurality of data files;
obtain a second subset of the plurality of data files;
obtain content index data corresponding to the second subset of the plurality of data files;
for each data file in the second subset of the plurality of data files, tag the respective data file using at least a portion of the content index data; and
classify the tagged data files in the second subset of the plurality of data files using the multi-class classifier.