Generating rules for classifying structured documents

A method for classifying a plurality of electronic structured documents to which a same schema is applied, the method comprising a computer performing steps of:

• determining one or more variable portions defined by the schema by scanning the schema;
• acquiring respective feature values of the determined variable portions from a structured document to be classified, wherein acquiring comprises reading content of the structured document while checking the schema;
• and
• classifying the structured document including the acquired feature values by applying the acquired feature values to rules,
• wherein the rules are rules for determining, on the basis of the feature values of the variable portions of the structured document to be classified, to which set of clustered structured documents the structured document to be classified belongs, wherein:
• the step of determining one or more variable portions includes the step of determining one or more elements included in a tree structure defined by the schema;
• the method further includes the step of acquiring respective feature values of the determined elements from the structured document;
• and
• each of the feature values of the determined elements is a number of repetitions of an element included in the tree structure, except when an element is included in at least two nodes belonging to a same definition on the schema, each of the feature values of the determined elements is an average of respective numbers of repetitions of the element in the at least two nodes.

A method for retrieving a structured document similar to a specific structured document from a plurality of electronic structured documents to which a same schema is applied, the method comprising a computer performing steps of:

• determining one or more variable portions defined by the schema by scanning the schema;
• acquiring respective feature values of the determined variable portions from the specific structured document and acquiring a first result by applying the acquired feature values to rules, wherein acquiring comprises reading content of the specific structured document while checking the schema;
• acquiring respective feature values of the determined variable portions from each of the plurality of structured documents to be searched and acquiring a second result by applying the acquired feature values to the rules for the structured document;
• and
• extracting a structured document similar to the specific structured document by comparing, for the structured document, a corresponding one of the second results with the first result,
• wherein the rules are rules for determining, on the basis of the feature values of the variable portions of the specific structured document or the structured document to be searched, to which set of clustered structured documents the specific structured document or the structured document to be searched belongs,
• wherein the step of determining one or more variable portions includes the step of determining one or more elements included in a tree structure defined by the schema;
• wherein the method further includes the step of acquiring respective feature values of the determined elements from the specific structured document and from each of the plurality of structured documents to be searched;
• and
• wherein each of the feature values of the determined elements is a number of repetitions of an element included in the tree structure, except when an element is included in at least two nodes belonging to a same definition on the schema, each of the feature values of the determined elements is an average of respective numbers of repetitions of the element in the at least two nodes.

A method for checking whether electronic structured documents to which a same schema is applied are similar to a specific structured document, the method comprising a computer performing steps of:

• determining one or more variable portions defined by the schema by scanning the schema;
• acquiring respective feature values of the determined variable portions from the specific structured document and acquiring a first result by applying the acquired feature values to rules, wherein acquiring comprises reading content of the specific structured document while checking the schema;
• acquiring respective feature values of the determined variable portions from a structured document to be checked and acquiring a second result by applying the acquired feature values to the rules;
• and
• checking, by comparing the second result with the first result, whether the structured document to be checked is similar to the specific structured document,
• wherein the rules are rules for determining, on the basis of the feature values of the variable portions of the specific structured document or the structured document to be checked, to which set of clustered structured documents the specific structured document or the structured document to be checked belongs,
• wherein the step of determining one or more variable portions includes the step of determining one or more elements included in a tree structure defined by the schema;
• wherein the method further includes the step of acquiring respective feature values of the determined elements from the specific structured document and the structured document to be checked;
• and
• wherein each of the feature values of the determined elements is a number of repetitions of an element included in the tree structure, except when an element is included in at least two nodes belonging to a same definition on the schema, each of the feature values of the determined elements is an average of respective numbers of repetitions of the element in the at least two nodes.