Identifying member profiles containing disallowed content in an online social network

1. A computer-implemented method for identifying member profiles containing disallowed content in an online social network, the method comprising:

scanning member profiles in the online social network, wherein each member profile includes a plurality of fields;

while scanning the member profiles, for each profile, looking up words from each of one or more fields in:

a set of multiple first dictionaries containing words that are correlated with disallowed profiles, and

at least one second dictionary containing words that are correlated with allowed profiles;

wherein the multiple first dictionaries are associated with different levels of disallowedness and contain words having differing frequencies of occurrence in the disallowed profiles; and

wherein each field in the plurality of fields is associated with a different set of multiple first dictionaries and a second dictionary;

computing a score for each member profile by:

determining counts of words in the member profile that are found in each of the multiple first dictionaries and the second dictionary associated with each field in the plurality of fields;

multiplying the count for each dictionary with an associated coefficient for the dictionary; and

totaling the multiplied counts for the plurality of fields to produce the score;

identifying one or more member profiles, wherein the member profile score of each identified member profile is above a threshold and/or among a highest N member profile scores (N>0); and

presenting the identified member profiles to a reviewer, wherein the reviewer verifies, for each identified member profile, whether the member profile contains disallowed content.

2. The computer-implemented method of claim 1 , wherein the presenting comprises:

displaying for the reviewer through an interface, for each identified member profile, an image and associated text from the member profile; and

allowing the reviewer to select and/or deselect each identified member profile through the interface, wherein a selection of a member profile by the reviewer verifies that the member profile contains disallowed content.

3. The computer-implemented method of claim 1 , wherein the associated coefficient for each dictionary is computed using a logistic-regression technique that examines words in a training set comprising member profiles that have been determined to be allowed profiles or disallowed profiles.

4. The computer-implemented method of claim 1 , wherein:

the online social network comprises an online professional network; and

the one or more fields in each member profile containing words that are looked up include:

first and last name fields containing a first name and a last name for the member;

a headline field specifying a job title and a company for the member; and

one or more position fields specifying prior positions held by the member.

5. The computer-implemented method of claim 1 , wherein the disallowed member profiles contain one or more of the following:

profanity;

pornography;

copyright violations;

trademark violations;

solicitations for sexual services;

advertisements for drugs; and

advertisements for counterfeit goods.

6. A non-transitory computer-readable storage medium storing instructions that, when executed by a computer, cause the computer to perform a method for identifying member profiles containing disallowed content in an online social network, the method comprising:

scanning member profiles in the online social network, wherein each member profile includes a plurality of fields;

while scanning the member profiles, for each profile, looking up words from each of one or more fields in:

a set of multiple first dictionaries containing words that are correlated with disallowed profiles, and

at least one second dictionary containing words that are correlated with allowed profiles;

wherein the multiple first dictionaries are associated with different levels of disallowedness and contain words having differing frequencies of occurrence in the disallowed profiles; and

wherein each field in the plurality of fields is associated with a different set of multiple first dictionaries and a second dictionary;

computing a score for each member profile by:

determining counts of words in the member profile that are found in each of the multiple first dictionaries and the second dictionary associated with each field in the plurality of fields;

multiplying the count for each dictionary with an associated coefficient for the dictionary; and

totaling the multiplied counts for the plurality of fields to produce the score;

identifying one or more member profiles, wherein the member profile score of each identified member profile is above a threshold and/or among a highest N member profile scores (N>0); and

presenting the identified member profiles to a reviewer, wherein the reviewer verifies, for each identified member profile, whether the member profile contains disallowed content.

7. The non-transitory computer-readable storage medium of claim 6 , wherein the presenting comprises:

displaying for the reviewing through an interface, for each identified member profile, an image and associated text from the member profile; and

allowing the reviewer to select and/or deselect each identified member profile through the interface, wherein a selection of a member profile by the reviewer verifies that the member profile contains disallowed content.

8. The non-transitory computer-readable storage medium of claim 6 , wherein the associated coefficient for each dictionary is computed using a logistic-regression technique that examines words in a training set comprising member profiles that have been determined to be allowed profiles or disallowed profiles.

9. The non-transitory computer-readable storage medium of claim 6 , wherein:

the online social network comprises an online professional network; and

the one or more fields in each member profile containing words that are looked up include:

first and last name fields containing a first name and a last name for the member;

a headline field specifying a job title and a company for the member; and

one or more position fields specifying prior positions held by the member.

10. The non-transitory computer-readable storage medium of claim 6 , wherein the disallowed member profiles contain one or more of the following:

profanity;

pornography;

copyright violations;

trademark violations;

solicitations for sexual services;

advertisements for drugs; and

advertisements for counterfeit goods.

11. A system that identifies member profiles containing disallowed content in an online social network, the system comprising:

a computing system comprising a processor and a memory;

wherein the computing system is configured to:

scan member profiles in the online social network, wherein each member profile includes a plurality of fields;

while scanning the member profiles, for each profile, look up words from each of one or more fields in:

a set of multiple first dictionaries containing words that are correlated with disallowed profiles, and

at least one second dictionary containing words that are correlated with allowed profiles;

wherein the multiple first dictionaries are associated with different levels of disallowedness and contain words having differing frequencies of occurrence in the disallowed profiles; and

wherein each field in the plurality of fields is associated with a different set of multiple first dictionaries and a second dictionary;

compute a score for each member profile by:

determining counts of words in the member profile that are found in each of the multiple first dictionaries and the second dictionary associated with each field in the plurality of fields;

multiplying the count for each dictionary with an associated coefficient for the dictionary; and

totaling the multiplied counts for the plurality of fields to produce the score; and

identify one or more member profiles, wherein the member profile score of each identified member profile is above a threshold and/or among a highest N member profile scores (N>0); and

present the identified member profiles to a reviewer, wherein the reviewer verifies, for each identified member profile, whether the member profile contains disallowed content.

12. The system of claim 11 , wherein the presenting comprises:

displaying for the reviewer through an interface, for each identified member profile, an image and associated text from the member profile; and

allowing the reviewer to select and/or deselect each identified member profile through the interface, wherein a selection of a member profile by the reviewer verifies that the member profile contains disallowed content.

13. The system of claim 11 , wherein the associated coefficient for each dictionary is computed using a logistic-regression technique that examines words in a training set comprising member profiles that have been determined to be allowed profiles or disallowed profiles.

14. The system of claim 11 , wherein:

the online social network comprises an online professional network; and

the one or more fields in each member profile containing words that are looked up includes:

first and last name fields containing a first name and a last name for the member;

a headline field specifying a job title and a company for the member; and

one or more position fields specifying prior positions held by the member.

15. The system of claim 11 , wherein the disallowed member profiles contain one or more of the following:

profanity;

pornography;

copyright violations;

trademark violations;

solicitations for sexual services;

advertisements for drags; and

advertisements for counterfeit goods.

16. The computer-implemented method of claim 1 , wherein a first set of multiple first dictionaries and a second dictionary is used for identifying at least one word in a first field of a first member profile as being allowed, while the one word in a second field of the first member profile is identified by a second set of multiple first dictionaries as being disallowed.

17. The computer-implemented method of claim 1 , further comprising creating the set of multiple first dictionaries by:

categorizing words into each dictionary based on predetermined percentages of occurrences of words in disallowed profiles and a minimum number of instances, wherein the words are taken from a training set comprising member profiles that have been determined to be disallowed profiles.

18. The non-transitory computer-readable storage medium of claim 6 , wherein a first set of multiple first dictionaries and a second dictionary is used for identifying at least one word in a first field of a first member profile as being allowed, while the one word in a second field of the first member profile is identified by a second set of multiple first dictionaries as being disallowed.

19. The non-transitory computer-readable storage medium of claim 6 , further comprising creating the set of multiple first dictionaries by:

categorizing words into each dictionary based on predetermined percentages of occurrences of words in disallowed profiles and a minimum number of instances, wherein the words are taken from a training set comprising member profiles that have been determined to be disallowed profiles.

20. The system of claim 11 , wherein a first set of multiple first dictionaries and a second dictionary is used for identifying at least one word in a first field of a first member profile as being allowed, while the one word in a second field of the first member profile is identified by a second set of multiple first dictionaries as being disallowed.

21. The system of claim 11 , further comprising creating the set of multiple first dictionaries by:

categorizing words into each dictionary based on predetermined percentages of occurrences of words in disallowed profiles and a minimum number of instances, wherein the words are taken from a training set comprising member profiles that have been determined to be disallowed profiles.