Data classifier
A document classifier may analyze documents for a search engine and tag the documents. A document classifier system may have several different classifiers, each with a separate algorithm for classification. Some of the data classifiers may learn or change the classification over time with a feedback loop. As those classifiers are modified, updated, replaced, or added, the documents that have already been classified by the classifier may be re-examined to update their classification. The document classifier system may maintain a database of documents with a timestamp that the document was classified that may be used to identify those documents whose classifications may be out of date.
1. A method performed on a computer processor, said method comprising:
receiving a first document having a document identifier;
performing a first analysis of said first document with a first classifier that identifies a first content item and tags said first content item with a first tag to create a first modified document comprising said first tag;
storing said document identifier with a document timestamp of said first analysis in a database;
determining that said first classifier has been modified to a second classifier and determining a classifier timestamp representing when said first classifier has been modified;
comparing said classifier timestamp to said document timestamp to determine that said classifier timestamp is newer than said document timestamp and performing a relook method comprising:
performing a second analysis of said first document with said second classifier that identifies a second content item and tags said second content item with a second tag to create a second modified document comprising said second tag; and
storing said document identifier with a second document timestamp of said second analysis in said database.
2. The method of claim 1 , said relook method comprising:
retrieving said first document from a document source.
3. The method of claim 1 , said relook method comprising:
retrieving said first document from a cache comprising a cached version of said first document.
4. The method of claim 1 , said relook method comprising:
said first document being said first modified document.
5. The method of claim 1 , said first classifier being trained using a feedback mechanism to create said second classifier.
6. The method of claim 1 further comprising:
adding a subset to said first classifier to create said second classifier.
7. The method of claim 6 , said second analysis comprising processing said first modified document using said subset.
8. A method performed on a computer processor, said method comprising:
receiving a first document having a document identifier;
performing a first analysis of said first document with a first classifier that identifies a first content item and tags said first content item with a first tag to create a first modified document comprising said first tag;
transmitting said modified first document to a classification system that classifies said first document using a first tag;
storing said document identifier with a document timestamp of said first analysis in a database;
determining that said first classifier has been modified to a second classifier and determining a classifier timestamp representing when said first classifier has been modified;
comparing said classifier timestamp to said document timestamp to determine that said classifier timestamp is newer than said document timestamp and performing a relook method comprising:
performing a second analysis of said first document with said second classifier that identifies a second content item and tags said second content item with a second tag to create a second modified document comprising said second tag;
transmitting said second modified document to said classification system; and
storing said document identifier with a second document timestamp of said second analysis in said database.
9. The method of claim 8 , said first classifier having a feedback mechanism and an adaptation mechanism that creates said second classifier.
10. The method of claim 9 , said feedback mechanism comprising input from said classification system.
11. The method of claim 10 , said classification system further comprising a search system.