Methods and systems for categorizing data in an on-demand database environment
View Patent ↗Categorizing data in an on-demand database environment is provided. The categorized data is accessed to provide results based on statistical likelihood that records provide a desired result of a query. The categorization of the data includes organizing queries based on semantic terms, with categorization based on a multidimensional categorization of data in the database environment. The generating of results includes accessing relationship metadata both for individual records and for categories. Relationships along the same category, or among categories can provide records that may answer the query. The relationships and statistics are updated based on usage of the results data. Records and relationships identified as being used to solve the query, or being a desired solution to the query, can be weighted more heavily, thus increasing the likelihood of providing the most relevant data for subsequent queries.
1. A system comprising:
a database system including hardware and software components to implement a multitenant database system (MTS), wherein the MTS stores data for multiple client organizations each identified by a tenant identifier (ID) and one or more users are associated with the tenant ID, wherein the one or more users of each client organization access data identified by the tenant ID associated with the respective client organization, and wherein the multitenant database is hosted by an entity separate from the client organization;
a server implemented on hardware components of the database system to:
receive a query from a client device associated with the tenant ID of the user of the client device;
categorize semantic terms contained in the query into one or more categories based on a previously stored multidimensional categorization scheme associated with the user's tenant ID;
access relationship metadata of records in the MTS previously categorized into one or more of the same categories as the query, the relationship metadata including first and second relationship types, wherein:
the first relationship type metadata indicate a record relationship between one record in the MTS to another record in the MTS, and
the second relationship type metadata indicate a category relationship between one of the categories of the one record in the MTS and other available categories of the multidimensional categorization scheme associated with the user's tenant ID, the category relationship based on a previously determined statistical similarity of the categories;
retrieve records from the MTS:
previously categorized into one or more of the same categories as the query,
having record relationships with any of the already retrieved records, and
previously categorized in a category having category relationships with one or more categories of any of the already retrieved records;
compute a statistical likelihood that each retrieved record is a desired solution to the query meriting inclusion in a result set for the query the statistical likelihood computed based on usage statistics accumulated for each retrieved record, including usage statistics accumulated for the record relationships and category relationships to other retrieved records;
monitor usage of the result set, including whether and for how long a record included in the result set is accessed, with greater access associated with solutions of greater desirability and lesser access associated with solutions of lesser desirability; and
update usage statistics indicating statistical similarity of records based on usage of the result set, including weighting more heavily records, record relationships and category relationships identified as being a desired solution to the query.
2. The system of claim 1 , further comprising the server to
store the query as a new record in the MTS, based on the previously stored multidimensional categorization scheme associated with the user's tenant ID.
3. The system of claim 1 , further comprising the server to
exclude records from a result set based on the statistical likelihood computed for each retrieved record whether it is a desired solution to the query.
4. The system of claim 1 , further comprising the server to cluster data based on the record relationships.
5. The system of claim 1 , further comprising the server to
send the query to a human agent if none of the retrieved records is used as a solution to the query; and
update the statistics responsive to a solution provided by the human agent.
6. A method comprising:
receiving a query at a multitenant database system (MTS), wherein the MTS stores data for multiple client organizations each identified by a tenant identifier (ID) and one or more users are associated with the tenant ID, wherein the one or more users of each client organization access data identified by the tenant ID associated with the respective client organization, and wherein the multitenant database is hosted by an entity separate from the client organization;
categorizing the semantic terms of the query into one or more categories based on a multidimensional categorization scheme stored in the MTS, the multidimensional categorization scheme associated with a tenant ID of a user that originated the query;
accessing relationship metadata of records in the MTS previously categorized into one or more of the same categories as the query, including accessing first and second relationship types, wherein:
the first relationship type metadata indicate a record relationship between one record in the MTS to another record in the MTS, and
the second relationship type metadata indicate a category relationship between one of the categories of the one record in the MTS and other available categories of the multidimensional categorization scheme associated with the user's tenant ID, the category relationship based on a previously determined statistical similarity of the categories;
retrieving records from the MTS:
previously categorized into one or more of the same categories as the query,
having record relationships with any of the already retrieved records, and
previously categorized in a category having category relationships with one or more categories of any of the already retrieved records;
computing a statistical likelihood that each retrieved record is a desired solution to the query meriting inclusion in a result set for the query, the statistical likelihood computed based on usage statistics accumulated for each retrieved record, including usage statistics accumulate for the record relationships and category relationships to other retrieved records;
monitoring usage of the result set, including whether and for how long a record included in the result set is used, with greater usage associated with solutions of greater desirability and lesser usage associated with solutions of lesser desirability; and
updating usage statistics indicating statistical similarity of records based on usage of the result set, including weighting more heavily records, record relationships and category relationships identified as being a desired solution to the query.
7. The method of claim 6 , wherein categorizing the semantic terms of the query comprises:
parsing the query into semantic terms; and
matching the parsed semantic terms to categories stored in the multidimensional categorization scheme associated with the tenant ID.
8. The method of claim 6 , wherein accessing relationship metadata further includes accessing relationship metadata of a record whose usage was monitored as being associated with a solution to multiple different queries, each query having been categorized into one or more categories of the multidimensional categorization scheme.
9. The method of claim 6 , further comprising:
storing the query as a new record in the MTS, based on the stored multidimensional categorization scheme associated with the user's tenant ID.
10. The method of claim 6 , further comprising:
excluding records from a result set based on the statistical likelihood computed for each retrieved record whether it is a desired solution to the query.
11. The method of claim 6 , further comprising:
clustering data based on the record relationships.
12. The method of claim 6 , further comprising:
sending the query to a human agent if none of the retrieved records is used as a solution to the query; and
updating the statistics responsive to a solution provided by the human agent.
13. An article of manufacture comprising a computer readable storage medium having content stored thereon, which when executed, cause a machine to perform operations including:
receiving a query at a multitenant database system (MTS), wherein the MTS stores data for multiple client organizations each identified by a tenant identifier (ID) and one or more users are associated with the tenant ID, wherein the one or more users of each client organization access data identified by the tenant ID associated with the respective client organization, and wherein the multitenant database is hosted by an entity separate from the client organization;
categorizing the semantic terms of the query into one or more categories based on a multidimensional categorization scheme stored in the MTS, the multidimensional categorization scheme associated with a tenant ID of a user that originated the query;
accessing relationship metadata of records in the MTS previously categorized into one or more of the same categories as the query, including accessing first and second relationship types, wherein:
the first relationship type metadata indicate a record relationship between one record in the MTS to another record in the MTS, and
the second relationship type metadata indicate a category relationship between one of the categories of the one record in the MTS and other available categories of the multidimensional categorization scheme associated with the user's tenant ID, the category relationship based on a previously determined statistical similarity of the categories;
retrieving records from the MTS:
previously categorized into one or more of the same categories as the query,
having record relationships with any of the already retrieved records, and
previously categorized in a category having category relationships with one or more categories of any of the already retrieved records;
computing a statistical likelihood that each retrieved record is a desired solution to the query meriting inclusion in a result set for the query, the statistical likelihood computed based on usage statistics accumulated for each retrieved record, including usage statistics accumulate for the record relationships and category relationships to other retrieved records;
monitoring usage of the result set, including whether and for how long a record included in the result set is used, with greater usage associated with solutions of greater desirability and lesser usage associated with solutions of lesser desirability; and
updating usage statistics indicating statistical similarity of records based on usage of the result set, including weighting more heavily records, record relationships and category relationships identified as being a desired solution to the query.
14. The article of manufacture of claim 13 , wherein the content to provide instructions for categorizing the semantic terms of the query comprises content to provide instructions for
parsing the query into semantic terms; and
matching the parsed semantic terms to categories stored in the multidimensional categorization scheme associated with the tenant ID.
15. The article of manufacture of claim 13 , further comprising content to provide instructions for
storing the query as a new record in the MTS, based on the stored multidimensional categorization scheme associated with the user's tenant ID.
16. The article of manufacture of claim 13 , further comprising content to provide instructions for
excluding records from a result set based on the statistical likelihood computed for each retrieved record whether it is a desired solution to the query.
17. The article of manufacture of claim 13 , further comprising content to provide instructions for
clustering data based on the record relationships.
18. The article of manufacture of claim 13 , further comprising content to provide instructions for
sending the query to a human agent if none of the retrieved records is used as a solution to the query; and
updating the statistics responsive to a solution provided by the human agent.
19. The system of claim 1 , wherein the usage statistics upon which the statistical likelihood that the retrieved record is a desired solution to the query is based are uniquely accumulated for each tenant ID.
20. The method of claim 6 , wherein the usage statistics upon which the statistical likelihood that the retrieved record is a desired solution to the query is based are uniquely accumulated for each tenant ID.
21. The article of manufacture of claim 13 , wherein the usage statistics upon which the statistical likelihood that the retrieved record is a desired solution to the query is based are uniquely accumulated for each tenant ID.