Constructing a domain-specific ontology by mining the web

Abstract

A method, apparatus, and program product is provided for constructing a domain-specific ontology for interpreting a policy. A phrase assumed to be associated with a policy is received. An indefinite term in the phrase is identified. Internet searches are performed on a plurality of terms from the phrase. A plurality of potential replacement terms is extracted from the Internet search results for the indefinite term. A context-specific ontology is built for the indefinite term based upon tracked occurrences of the plurality of potential replacement terms. The policy is interpreted by accessing the domain-specific ontology to interpret the indefinite term, where the indefinite term is interpreted by mapping the indefinite term to a plurality of potential replacement terms in the ontology for a plurality of contexts, and creating a policy based upon the interpretation of the indefinite term from the ontology.

1 . A method for interpreting a policy, the method comprising:
 receiving a phrase assumed to be associated with a policy, wherein the policy includes a subject term, an action term, and a resource term;   identifying a component of the policy in the phrase;   identifying an indefinite term in the component of the policy by consulting a dictionary, wherein identifying the indefinite term includes identifying that the indefinite term is indefinite as a result of having different potential definitions in different contexts, and wherein the indefinite term is a subject term, an action term, or a resource term;   accessing a domain-specific ontology to interpret the indefinite term, wherein the indefinite term is interpreted by mapping the indefinite term to at least one selected potential replacement term from among a plurality of potential replacement terms in the ontology for a plurality of contexts, wherein the plurality of potential replacement terms are associated with different definitions of the indefinite term; and   creating a policy based upon the interpretation of the indefinite term from the ontology by substituting the indefinite term with the at least one selected potential replacement term in the policy.   
     
     
         2 . The method of  claim 1  wherein the policy is a security policy. 
     
     
         3 . The method of  claim 1  wherein the components of the policy include a subject, an action, and a resource. 
     
     
         4 . The method of  claim 1  wherein accessing the domain-specific ontology comprises:
 checking the context of the indefinite term; 
 returning one or more terms from the domain-specific ontology that map to the indefinite term if the indefinite term is in the context of the domain-specific ontology; and 
 expanding the domain-specific ontology to include ontology in a given context if the indefinite term is not in the context of the domain-specific ontology. 
 
     
     
         5 . The method of  claim 4  wherein expanding the domain-specific ontology comprises:
 performing a first Internet search on a plurality of terms from the phrase to generate first search results; 
 extracting from the first search results a plurality of clues associated with potential definitions of the indefinite term; 
 performing a second Internet search incorporating the plurality of clues to generate second search results; 
 extracting from the second search results a plurality of potential replacement terms of the indefinite term; 
 tracking occurrences of each of the plurality of potential replacement terms in the second search results; and 
 updating the context-specific ontology for the indefinite term based upon the tracked occurrences of the plurality of potential replacement terms. 
 
     
     
         6 . The method of  claim 5  further comprising:
 restricting the first search results to a domain of interest by removing hits received from the first Internet search from the first search results that are unrelated to the domain of interest. 
 
     
     
         7 . The method of  claim 1  wherein interpreting the indefinite term comprises:
 using the domain-specific ontology to statistically determine a plurality of terms to which the indefinite term can map in a particular context; 
 evaluating a probability associated with each of the plurality of terms; and 
 selecting one or more terms from the plurality of terms based upon the probability. 
 
     
     
         8 . An apparatus comprising:
 a processor;   a memory; and   program code configured to be executed by the processor for interpreting a policy, the program code resident in the memory and configured to:
 receive a phrase assumed to be associated with a policy, wherein the policy includes a subject term, an action term, and a resource term, 
 identify a component of the policy in the phrase, 
 identify an indefinite term in the component of the policy by consulting a dictionary, wherein the program code is configured to identify the indefinite term by identifying that the indefinite term is indefinite as a result of having different potential definitions in different contexts, and wherein the indefinite term is a subject term, an action term, or a resource term, 
 access a domain-specific ontology to interpret the indefinite term where the indefinite term is interpreted by mapping the indefinite term to at least one selected potential replacement term from among a plurality of potential replacement terms in the ontology for a plurality of contexts, wherein the plurality of potential replacement terms are associated with different definitions of the indefinite term, and 
 create a policy based upon the interpretation of the indefinite term from the ontology by substituting the indefinite term with the at least one selected potential replacement term in the policy. 
   
     
     
         9 . The apparatus of  claim 8  wherein the components of the policy include a subject, an action, and a resource. 
     
     
         10 . The apparatus of  claim 8  wherein the program code is configured to access the domain-specific ontology by:
 checking the context of the indefinite term; 
 returning one or more terms from the domain-specific ontology that map to the indefinite term if the indefinite term is in the context of the domain-specific ontology; and 
 expanding the domain-specific ontology to include ontology in a given context if the indefinite term is not in the context of the domain-specific ontology. 
 
     
     
         11 . The apparatus of  claim 10  wherein the program code is configured to expand the domain-specific ontology by:
 performing a first Internet search on a plurality of terms from the phrase to generate first search results; 
 extracting from the first search results a plurality of clues associated with potential definitions of the indefinite term; 
 performing a second Internet search incorporating the plurality of clues to generate second search results; 
 extracting from the second search results a plurality of potential replacement terms of the indefinite term; 
 tracking occurrences of each of the plurality of potential replacement terms in the second search results; and 
 updating the context-specific ontology for the indefinite term based upon the tracked occurrences of the plurality of potential replacement terms. 
 
     
     
         12 . The apparatus of  claim 11  wherein the program code is further configured to restrict the first search results to a domain of interest by removing hits received from the first Internet search from the first search results that are unrelated to the domain of interest. 
     
     
         13 . The apparatus of  claim 8  wherein the program code is configured to interpret the indefinite term by:
 using the domain-specific ontology to statistically determine a plurality of terms to which the indefinite term can map in a particular context; 
 evaluating a probability associated with each of the plurality of terms; and 
 selecting one or more terms from the plurality of terms based upon the probability. 
 
     
     
         14 . A program product, comprising:
 a recordable type computer readable medium; and   program code configured to interpret a policy, the program code resident on the recordable type computer readable medium and configured to:
 receive a phrase assumed to be associated with a policy, wherein the policy includes a subject term, an action term, and a resource term, 
 identify a component of the policy in the phrase, 
 identify an indefinite term in the component of the policy by consulting a dictionary, wherein the program code is configured to identify the indefinite term by identifying that the indefinite term is indefinite as a result of having different potential definitions in different contexts, and wherein the indefinite term is a subject term, an action term, or a resource term, 
 access a domain-specific ontology to interpret the indefinite term where the indefinite term is interpreted by mapping the indefinite term to at least one selected potential replacement term from among a plurality of potential replacement terms in the ontology for a plurality of contexts, wherein the plurality of potential replacement terms are associated with different definitions of the indefinite term, and 
 create a policy based upon the interpretation of the indefinite term from the ontology by substituting the indefinite term with the at least one selected potential replacement term in the policy. 
   
     
     
         15 . The program product of  claim 14  wherein the components of the policy include a subject, an action, and a resource. 
     
     
         16 . The program product of  claim 14  wherein the program code is configured to access the domain-specific ontology by:
 checking the context of the indefinite term; 
 returning one or more terms from the domain-specific ontology that map to the indefinite term if the indefinite term is in the context of the domain-specific ontology; and 
 expanding the domain-specific ontology to include ontology in a given context if the indefinite term is not in the context of the domain-specific ontology. 
 
     
     
         17 . The program product of  claim 16  wherein the program code is configured to expand the domain-specific ontology by:
 performing a first Internet search on a plurality of terms from the phrase to generate first search results; 
 extracting from the first search results a plurality of clues associated with potential definitions of the indefinite term; 
 performing a second Internet search incorporating the plurality of clues to generate second search results; 
 extracting from the second search results a plurality of potential replacement terms of the indefinite term; 
 tracking occurrences of each of the plurality of potential replacement terms in the second search results; and 
 updating the context-specific ontology for the indefinite term based upon the tracked occurrences of the plurality of potential replacement terms. 
 
     
     
         18 . The program product of  claim 17  wherein the program code is further configured to restrict the first search results to a domain of interest by removing hits received from the first Internet search from the first search results that are unrelated to the domain of interest. 
     
     
         19 . The program product of  claim 14  wherein the program code is configured to interpret the indefinite term by:
 using the domain-specific ontology to statistically determine a plurality of terms to which the indefinite term can map in a particular context; 
 evaluating a probability associated with each of the plurality of terms; and selecting one or more terms from the plurality of terms based upon the probability.