System and method for automated recommendation generation from communication content and application thereof

1. A method implemented on at least one processor, a memory, and a communication platform for product recommendation, comprising:

receiving communication content from a plurality of service providers each operating on a respective one of a plurality of different platforms;

identifying automatically, via machine learning, information related to a product mentioned on a webpage by:

extracting an HTML link to a webpage included in the communication content,

accessing the webpage using the extracted HTML link, wherein the webpage includes different contents organized in accordance with a specified structure including an HTML tree with a plurality of nodes, and

extracting a name of the product using a product name classification model obtained via machine learning and a plurality of other pieces of information relevant to the product from the webpage based on the specified structure;

identifying, based on machine-learned demographics/interests/platforms (DIP) targeting models, a plurality of targets associated with each of the one or more service providers as an intended targeted audience for a recommendation of the product;

generating, for each of the plurality of different platforms, a corresponding recommendation of the product based on the information related to the product, wherein generating the corresponding recommendation of the product for each of the plurality of different platforms comprises including the extracted name of the product and a partial subset of the extracted plurality of other pieces of information relevant to the product in each corresponding recommendation of the product;

formatting each recommendation of the product in a form appropriate with respect to the service provider on the corresponding platform;

sending, to each of the plurality of service providers, the corresponding recommendation of the product with targeting instructions indicating the intended targeted audience for distributing the recommendation of the product to the intended targeted audience; and

updating, based on user activities associated with the product received from the one or more service providers, the demographics/interests/platforms (DIP) targeting models.

2. The method of claim 1 , wherein the communication content includes at least one of:

an electronic mail involving a user;

a short message associated with a user; and

textual information presented in a chatroom with a group of users.

3. The method of claim 1 , wherein the plurality of nodes include a root node, leave nodes, and intermediate nodes, wherein each of the intermediate nodes represents a sub-tree with nodes from the plurality of nodes originated from the intermediate node.

4. The method of claim 3 , wherein the step of extracting the name of the product comprises:

identifying one of the intermediate nodes in the HTML tree corresponding to a sub-tree including a single link therein;

determining a link-based boundary within the webpage defined by the identified sub-tree;

generating a plurality of phrases based on text of the webpage within the link-based boundary; and

selecting one of the plurality of phrases as the name of the product via classification performed based on the product name classification model.

5. The method of claim 4 , further comprising:

filtering, prior to the step of selecting, the plurality of phrases to remove those phrases that are not corresponding to a name of a product; and

obtaining additional information related to the product, wherein

the additional information related to the product includes a visual presentation of the product and a hypertext link through which further information associated with the product can be retrieved, and

the additional information is extracted from either the webpage within the link-based boundary or from content at another source linked via the hypertext link that is present on the webpage within the link-based boundary.

6. The method of claim 1 , wherein the step of identifying the plurality of targets comprises:

accessing the DIP targeting models created based on tracked information associated with users of the plurality of service providers operating on the plurality of platforms; and

identifying, based on the DIP targeting models, the plurality of targets as intended targeted audience for the recommendation of the product with respect to each of the plurality of platforms, wherein

the tracked information records activities of the users with respect to different types of information including previously generated product recommendations distributed thereto.

7. Machine readable and non-transitory medium having information recorded thereon for product recommendation, wherein the information, when read by a machine, causes the machine to perform following steps:

receiving communication content from a plurality of service providers each operating on a respective one of a plurality of different platforms;

identifying automatically, via machine learning, information related to a product mentioned on a webpage by:

extracting an HTML link to a webpage included in the communication content,

accessing the webpage using the extracted HTML link, wherein the webpage includes different contents organized in accordance with a specified structure including an HTML tree with a plurality of nodes, and

extracting a name of the product using a product name classification model obtained via machine learning and a plurality of other pieces of information relevant to the product from the webpage based on the specified structure;

identifying, based on machine-learned demographics/interests/platforms (DIP) targeting models, a plurality of targets associated with each of the one or more service providers as an intended targeted audience for a recommendation of the product;

generating, for each of the plurality of different platforms, a corresponding recommendation of the product based on the information related to the product, wherein generating the corresponding recommendation of the product for each of the plurality of different platforms comprises including the extracted name of the product and a partial subset of the extracted plurality of other pieces of information relevant to the product in each corresponding recommendation of the product;

formatting each recommendation of the product in a form appropriate with respect to the service provider on the corresponding platform;

sending, to each of the plurality of service providers, the corresponding recommendation of the product with targeting instructions indicating the intended targeted audience for distributing the recommendation of the product to the intended targeted audience; and

updating, based on user activities associated with the product received from the one or more service providers, the demographics/interests/platforms (DIP) targeting models.

8. The medium of claim 7 , wherein the communication content includes at least one of:

an electronic mail involving a user;

a short message associated with a user; and

textual information presented in a chatroom with a group of users.

9. The medium of claim 7 , wherein

the plurality of nodes include a root node, leave nodes, and intermediate nodes, wherein each of the intermediate nodes represents a sub-tree with nodes from the plurality of nodes originated from the intermediate node.

10. The medium of claim 9 , wherein the step of extracting the name of the product comprises:

identifying one of the intermediate nodes in the HTML tree corresponding to a sub-tree including a single link therein;

determining a link-based boundary within the webpage defined by the identified sub-tree;

generating a plurality of phrases based on text of the webpage within the link-based boundary; and

selecting one of the plurality of phrases as the name of the product via classification performed based on the product name classification model.

11. The medium of claim 10 , wherein the information, when read by the machine, further causes the machine to perform the following steps:

filtering, prior to the step of selecting, the plurality of phrases to remove those phrases that are not corresponding to a name of a product; and

obtaining additional information related to the product, wherein

the additional information related to the product includes a visual presentation of the product and a hypertext link through which further information associated with the product can be retrieved, and

the additional information is extracted from either the webpage within the link-based boundary or from content at another source linked via the hypertext link that is present on the webpage within the link-based boundary.

12. The medium of claim 7 , wherein the step of identifying the plurality of targets comprises:

accessing the DIP targeting models created based on tracked information associated with users of the plurality of service providers operating on the plurality of platforms; and

identifying, based on the DIP targeting models, the plurality of targets as intended targeted audience for the recommendation of the product with respect to each of the plurality of platforms, wherein

the tracked information records activities of the users with respect to different types of information including previously generated product recommendations distributed thereto.

13. A system for product recommendation, comprising:

a product recommendation generator implemented by a processor and configured for

receiving communication content from a plurality of service providers each operating on a corresponding respective one of a plurality of different platforms;

identifying automatically, via machine learning, information related to a product mentioned on a webpage by:

extracting an HTML link to a webpage included in the communication content,

accessing the webpage using the extracted HTML link, wherein the webpage includes different contents organized in accordance with a specified structure including an HTML tree with a plurality of nodes, and

extracting a name of the product using a product name classification model obtained via machine learning and a plurality of other pieces of information relevant to the product from the webpage based on the specified structure;

identifying, based on machine-learned demographics/interests/platforms (DIP) targeting models, a plurality of targets associated with each of the one or more service providers as an intended targeted audience for a recommendation of the product; and

generating, for each of the plurality of different platforms, a corresponding recommendation of the product based on the information related to the product, wherein generating the corresponding recommendation of the product for each of the plurality of different platforms comprises including the extracted name of the product and a partial subset of the extracted plurality of other pieces of information relevant to the product in each corresponding recommendation of the product; and

a recommendation card distribution engine implemented by a processor and configured for

formatting each recommendation of the product in a form appropriate with respect to the service provider on the corresponding platform;

sending, to each of the plurality of service providers, the corresponding recommendation of the product with targeting instructions indicating the intended targeted audience for distributing the recommendation of the product to the intended targeted audience; and

updating, based on user activities associated with the product received from the one or more service providers, the demographics/interests/platforms (DIP) targeting models.

14. The system of claim 13 , wherein the communication content includes at least one of:

an electronic mail involving a user;

a short message associated with a user; and

textual information presented in a chatroom with a group of users.

15. The system of claim 13 , wherein

the plurality of nodes include a root node, leave nodes, and intermediate nodes, wherein each of the intermediate nodes represents a sub-tree with nodes from the plurality of nodes originated from the intermediate node.

16. The system of claim 15 , wherein the step of extracting the name of the product comprises:

identifying one of the intermediate nodes in the HTML tree corresponding to a sub-tree including a single link therein;

determining a link-based boundary within the webpage defined by the identified sub-tree;

generating a plurality of phrases based on text of the webpage within the link-based boundary; and

selecting one of the plurality of phrases as the name of the product via classification performed based on the product name classification model.

17. The method of claim 16 , further comprising:

filtering, prior to the step of selecting, the plurality of phrases to remove those phrases that are not corresponding to a name of a product; and

obtaining additional information related to the product, wherein

the additional information related to the product includes a visual presentation of the product and a hypertext link through which further information associated with the product can be retrieved, and

the additional information is extracted from either the webpage within the link-based boundary or from content at another source linked via the hypertext link that is present on the webpage within the link-based boundary.