Recommender evaluation based on tokenized messages
View Patent ↗A machine may implement a recommender that provides recommendations to users. The machine may be configured to present a first version of the recommender configured by various parameters. A user may submit a message to the machine, and the machine may identify a parameter among the various parameters by tokenizing the message and identifying the parameter among the tokens. The machine may then generate a second version of the recommender by modifying the parameter and configuring the second version according to the modified parameter. The machine may then present the first and second versions of the recommender contemporaneously two different portions of the users. By tokenizing a further message received from the users, the machine may evaluate the first and second versions and determine whether the second version is a replacement of the first version.
1. A method comprising:
presenting a first version of a recommender configured to provide users with recommendations based on a plurality of parameters that configure the first version of the recommender;
identifying a parameter among the plurality of parameters by tokenizing a message submitted by a user among the users and identifying the parameter among tokens within the tokenized message;
generating a second version of the recommender by modifying the parameter identified among the tokens and configuring the second version according to the modified parameter,
the generating of the second version of the recommender being performed by a processor of a machine; and
presenting the first version of the recommender contemporaneously with the second version of the recommender,
the first version being presented to a first portion of the users,
the second version being presented to a second portion of the users.
2. The method of claim 1 further comprising:
tokenizing a further message submitted by a further user among the first portion of the users; and
determining that the second version of the recommender is a replacement of the first version of the recommender based on the parameter being among tokens of the tokenized further message.
3. The method of claim 1 further comprising:
tokenizing a further message submitted by a further user among the first portion of the users; and
determining that the second version of the recommender is not a replacement of the first version of the recommender based on the parameter being among tokens of the tokenized further message.
4. The method of claim 1 further comprising:
tokenizing a further message submitted by a further user among the second portion of the users; and
determining that the second version of the recommender is a replacement of the first version of the recommender based on the parameter being among tokens of the tokenized further message.
5. The method of claim 1 further comprising:
tokenizing a further message submitted by a further user among the second portion of the users; and
determining that the second version of the recommender is not a replacement of the first version of the recommender based on the parameter being among tokens of the tokenized further message.
6. The method of claim 1 , wherein:
the identifying of the parameter among the tokens includes determining a sentiment of the message submitted by the user; and
the modifying of the parameter is based on the sentiment of the message.
7. The method of claim 6 , wherein:
the determining of the sentiment determines that the sentiment is positive; and
the modifying of the parameter is based on the sentiment being positive.
8. The method of claim 6 , wherein:
the determining of the sentiment determines that the sentiment is negative; and
the modifying of the parameters is based on the sentiment being negative.
9. The method of claim 1 , wherein:
the recommendations refer to jobs recommended to the users; and
the parameter is a threshold distance according to which the first version of the recommender determines whether to provide a recommendation of a job to the user.
10. The method of claim 1 , wherein:
the recommendations refer to social contacts recommended to the users;
the parameter is a threshold age according to which the first version of the recommender determines whether to provide a recommendation of a social contact to the user.
11. The method of claim 1 , wherein:
the recommendations refer to romantic contacts recommended to the users;
the parameter is a threshold distance according to which the first version of the recommender determines whether to provide a recommendation of a romantic contact to the user.
12. The method of claim 1 , wherein:
the recommendations refer to products recommended to the users; and
the parameter is a threshold price according to which the first version of the recommender determines whether to provide a recommendation of a product to the user.
13. The method of claim 1 , wherein:
the tokenizing of the message includes tokenizing an email received from the user.
14. The method of claim 1 , wherein:
the tokenizing of the message includes tokenizing a blog post published on behalf of the user.
15. The method of claim 1 , wherein:
the identifying of the parameter identifies a unigram token among the tokens within the tokenized message as the parameter.
16. The method of claim 1 , wherein:
the identifying of the parameter identifies a bigram token among the tokens within the tokenized message as the parameter.
17. A non-transitory machine-readable storage medium comprising instructions that, when executed by one or more processors of a machine, cause the machine to perform operations comprising:
presenting a first version of a recommender configured to provide users with recommendations based on a plurality of parameters that configure the first version of the recommender;
identifying a parameter among the plurality of parameters by tokenizing a message submitted by a user among the users and identifying the parameter among tokens within the tokenized message;
generating a second version of the recommender by modifying the parameter identified among the tokens and configuring the second version according to the modified parameter,
the generating of the second version of the recommender being performed by the one or more processors of the machine; and
presenting the first version of the recommender contemporaneously with the second version of the recommender,
the first version being presented to a first portion of the users,
the second version being presented to a second portion of the users.
18. The non-transitory machine-readable storage medium of claim 17 , wherein the operations further comprise:
tokenizing a further message submitted by a further user among the second portion of the users; and
determining that the second version of the recommender is a replacement of the first version of the recommender based on the parameter being among tokens of the tokenized further message.
19. A system comprising:
a presentation module configured to present a first version of a recommender configured to provide users with recommendations based on a plurality of parameters that configure the first version of the recommender;
a feedback module configured to identify a parameter among the plurality of parameters by tokenizing a message submitted by a user among the users and identifying the parameter among tokens within the tokenized message;
a processor configured by a modification module to generate a second version of the recommender by modifying the parameter identified among the tokens and configuring the second version according to the modified parameter; and
an evaluation module configured to present the first version of the recommender contemporaneously with the second version of the recommender,
the first version being presented to a first portion of the users,
the second version being presented to a second portion of the users.
20. The system of claim 19 , wherein:
the recommendations refer to jobs recommended to the users; and
the parameter is a threshold distance according to which the first version of the recommender determines whether to provide a recommendation of a job to the user.