On-the fly pattern recognition with configurable bounds
View Patent ↗Some embodiments of on-the-fly pattern recognition with configurable bounds have been presented. In one embodiment, a pattern matching engine is configured based on user input, which may include values of one or more user configurable bounds on searching. Then the configured pattern matching engine is used to search for a set of features in an incoming string. A set of scores is updated based on the presence of any of the features in the string while searching for the features. Each score may indicate a likelihood of the content of the string being in a category. The search is terminated if the end of the string is reached or if the user configurable bounds are met. After terminating the search, the scores are output.
1. A model-based content classification system for use in a firewall, comprising:
a user interface that receives from a user one or more user-configurable bounds on searching content;
a pattern matching engine that receives the one or more user-configurable bounds from the user interface, wherein the pattern matching engine is stored in a memory associated with the firewall and wherein execution of the pattern matching engine by a processor associated with the firewall:
searches a string having known content, for a predetermined pattern, the search limited according to the received user-configurable bounds,
computes a plurality of scores indicating a likelihood that the content of the string corresponds to one or more predetermined content categories, and
updates the scores using a plurality of weights associated with the predetermined pattern in response to detecting the predetermined pattern;
a content classification model generator that receives the updated scores from the pattern matching engine, wherein the content classification model generator is stored in the memory associated with the firewall and wherein execution of the content classification model generator by the processor associated with the firewall:
generates a content classification model based on the updated scores, and
transmits the content classification model to a model repository stored in memory;
a network interface of the firewall that receives subsequent content over a network connection, the subsequent content directed to a client computer that is communicatively coupled to the firewall;
a content classification engine stored in the memory associated with the firewall, wherein execution of the content classification engine by the processor associated with the firewall:
accesses the content classification model stored in the model repository, and
classifies the subsequent content based on the content classification model;
a content filter stored in the memory associated with the firewall, wherein execution of the content filter by the processor associated with the firewall transmits at least a subset of the subsequent content to the client computer, the subset selected based on the classification of the subsequent content.
2. The model-based content classification system of claim 1 , wherein the content classification engine classifies subsequent content based on the content classification model by comparing a plurality of scores assigned to the subsequent content to the scores of the generated content classification model.
3. The model-based content classification system of claim 2 , wherein the content classification engine classifies subsequent content based on the content classification model by further determining that the subsequent content is prohibited based on a predetermined policy.
4. The model-based content classification system of claim 1 , wherein further execution of the pattern matching engine further terminates the search when the end of the string is reached or one or more of the configurable bounds are met.
5. The model-based content classification system of claim 1 , wherein the one or more configurable bounds includes a predetermined sample size.
6. The model-based content classification system of claim 1 , wherein the one or more configurable bounds includes a predetermined search time limit.
7. The model-based content classification system of claim 1 , wherein the one or more configurable bounds includes a predetermined threshold of a confidence level of the content of the string being in one of the plurality of categories.
8. The model-based content classification system of claim 1 , wherein the one or more configurable bounds includes a minimum salience.
9. The model-based content classification system of claim 1 , wherein the one or more configurable bounds includes a maximum barren sample size.
10. The model-based content classification system of claim 1 , wherein the content classification engine accesses the content classification model stored in the model repository without downloading the content classification model.
11. The model-based content classification system of claim 1 , wherein the content classification engine accesses the content classification model stored in the model repository by downloading the content classification model.
12. The model-based content classification of claim 1 , wherein the content classification model is a model that identifies pornographic content.
13. The model-based content classification of claim 1 , wherein the content classification model is a model that identifies spam.
14. The model-based content classification system of claim 1 , wherein the predetermined pattern includes a pre-selected token.
15. The model-based content classification system of claim 1 , wherein the predetermined pattern includes an N-gram.
16. The model-based content classification system of claim 1 , wherein the subsequent content is a part of a web page.
17. The model-based content classification system of claim 1 , wherein the subsequent content is a part of an email.