Automatic detection of online commercial intention

• US 7,831,685 B2
• Filed: 12/14/2005
• Issued: 11/09/2010
• Est. Priority Date: 12/14/2005
• Status: Expired due to Fees

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Nie, Zaiqing, Wen, Ji-Rong, Zhao, Lingzhi, Dai, Honghua, Li, Ying, Wang, Lee
• Primary Examiner(s): Bruckart; Benjamin R
• Assistant Examiner(s): Waqas; Saad A
• Application Number:
  US11/300,748
• Publication Number:
  US 20070136457A1
• Time in Patent Office: 1,791 Days
• Field of Search: 705/26, 705/1, 705/14, 705/80, 705/53, 705/51, 705/7, 705/57, 705/28, 707/3, 707/E17.11, 707/5, 707/100, 707/104.1, 709/217, 709/224
• US Class Current: 709/217
• CPC Class Codes: G06Q 30/02 Marketing; Price estimation or determination; Fundraising

Abstract

Features extracted from network browser pages and/or network search queries are leveraged to facilitate in detecting a user'"'"'s browsing and/or searching intent. Machine learning classifiers constructed from these features automatically detect a user'"'"'s online commercial intention (OCI). A user'"'"'s intention can be commercial or non-commercial, with commercial intentions being informational or transactional. In one instance, an OCI ranking mechanism is employed with a search engine to facilitate in providing search results that are ranked according to a user'"'"'s intention. This also provides a means to match purchasing advertisements with potential customers who are more than likely ready to make a purchase (transactional stage). Additionally, informational advertisements can be matched to users who are researching a potential purchase (informational stage).

Claims

1. One or more non-transitory, tangible computer storage media having a system embodied thereon including computer-executable instructions that, when executed, perform a method for determining a user'"'"'s purchasing intent, comprising:

a training component that;

receives at least one of a training web page having a first purchasing intention label or a training query having a second purchasing intention label,

wherein the first purchasing intention label identifies a pre-determined purchasing intent for the training web page and the second purchasing intention label identifies a pre-determined purchasing intent for the training query,

wherein the training web page is a web page publicly accessible via the Internet,

wherein the training query is an Internet search query,

wherein purchasing intent is a degree of interest, while viewing a web page or conducting a query, that a user has in purchasing one or more products or services related to the web page or query, and

wherein purchasing intent is classified as one of;

purchase, informational, or non-purchase, purchase indicating the user has a strong interest in purchasing a product related to the web page or query, informational indicating the user has moderate interest in purchasing and/or is researching, and non-purchase indicating the user has low or no interest in purchasing,

upon receiving a training web page, extracts one or more purchasing intent indicators from the training web page, purchasing intent indicators comprising web page content, keyword patterns, web page feature positions, links, and web page layout,

upon receiving a training query, inputs the training query to a search engine, receives a search results web page from the search engine including Internet web page links, each link having an associated Internet web page, and extracts one or more purchasing intent indicators from the search results web page and a plurality of the associated Internet web pages, and

trains a purchasing intent model by applying a machine learning algorithm based on at least one of;

the purchasing intent indicators extracted from the training web page and the first purchasing intention label, or

the purchasing intent indicators extracted from the search results web page of the training query or from the associated Internet web pages and the second purchasing intention label,

wherein when a training query is received, the purchasing intent model determines a purchasing intent for the search results web page and the plurality of the associated Internet web pages and determines combine coefficients for combining the purchasing intent of the search results web page and each of the plurality of Internet web pages such that the combination of purchasing intents corresponds to the pre-determined purchasing intent identified in the second purchasing intention label;

an intake component that receives at least one of user Internet search query information or user Internet web page browsing information; and

a classifier component that;

upon receiving user Internet search query information, inputs the user Internet search query information to a search engine, receives a search results web page from the search engine including Internet web page links, each link having at least one associated Internet web page, and extracts at least one purchasing intent indicator from the search results web page and a plurality of the associated Internet web pages,

upon receiving user Internet web page browsing information, extracts at least one purchasing intent indicator from the user Internet web page browsing information, and

determines a purchasing intent for the received at least one of user Internet search query information or user Internet web page browsing information by providing the extracted purchasing intent indicators to the trained purchasing intent model,

wherein when purchasing intent for received user Internet search query information is determined, a purchasing intent for the search results web page and the plurality of associated Internet web pages are determined, and the purchasing intent of the search results web page and each of the plurality of Internet web pages are combined according to the combine coefficients determined by the purchasing intent model to determine the purchasing intent for the received user Internet search query.

2. The media of claim 1, wherein the intake component receives at least one user Internet search query.

3. The media of claim 1, wherein the classifier component employs machine learning to determine purchasing intent.

4. The media of claim 1, wherein the system is employed by a search engine to rank search results based on purchasing intent.

11. The media of claim 1, wherein the system is implemented on at least one of a computer, a server, or a handheld electronic device.

5. One or more non-transitory, tangible computer storage media storing computer-executable instructions for performing a method for determining purchasing intent from network-based information, the method comprising:

obtaining at least one of a training web page having a first purchasing intention label or a training query having a second purchasing intention label,

wherein the first purchasing intention label identifies a pre-determined purchasing intent for the training web page and the second purchasing intention label identifies a pre-determined purchasing intent for the training query,

wherein the training web page is a web page publicly accessible via the Internet,

wherein the training query is an Internet search query,

wherein purchasing intent is a degree of interest, while viewing a web page or conducting a query, that a user has in purchasing one or more products or services related to the web page or query, and

wherein purchasing intent is classified as one of;

purchase, informational, or non-purchase, purchase indicating the user has a strong interest in purchasing a product related to the web page or query, informational indicating the user has moderate interest in purchasing and/or is researching, and non-purchase indicating the user has low or no interest in purchasing;

upon obtaining a training web page, extracting one or more purchasing intent indicators from the training web page, purchasing intent indicators comprising web page content, keyword patterns, web page feature positions, links, and web page layout;

upon receiving a training query, submitting the training query to a search engine, receiving a search results web page from the search engine including Internet web page links, each link having an associated Internet web page, and extracting one or more purchasing intent indicators from the search results web page and a plurality of the associated Internet web pages;

training a purchasing intent model by applying a machine learning algorithm based on at least one of;

the purchasing intent indicators extracted from the training web page and the first purchasing intention label, or

the purchasing intent indicators extracted from the search results web page of the training query or from the associated Internet web pages and the second purchasing intention label,

wherein when a training query is received, determining a purchasing intent for the search results web page and the plurality of associated Internet web pages and determining combine coefficients for combining the purchasing intent of the search results web page and each of the plurality of Internet web pages such that the combination of purchasing intents corresponds to the pre-determined purchasing intent identified in the second purchasing intention label;

receiving at least one of user Internet search query information or user Internet web page browsing information;

upon receiving user Internet search query information, submitting the user Internet search query information to a search engine, receiving a search results web page from the search engine including Internet web page links, each link having at least one associated Internet web page, and extracting one or more purchasing intent indicators from the search results web page and a plurality of the associated Internet web pages;

upon receiving user Internet web page browsing information, extracting at least one purchasing intent indicator from the user Internet web page browsing information; and

determining a purchasing intent for the received at least one of user Internet search query information or user Internet web page browsing information by providing the extracted purchasing intent indicators to the trained purchasing intent model,

wherein determining purchasing intent for received user Internet search query information comprises determining a purchasing intent for the search results web page and the plurality of the associated Internet web pages and combining the purchasing intent of the search results web page and each of the plurality of Internet web pages according to the combine coefficients determined by the purchasing intent model to determine the purchasing intent for the received user Internet search query.

6. The media of claim 5, further comprising automatically classifying Internet web pages according to purchasing intent.

7. The media of claim 5, wherein the purchasing intent for received user Internet search query information is determined using a machine learning algorithm.

8. The media of claim 5, wherein determination of purchasing intent for received user Internet search query information is done automatically.

9. The media of claim 5, wherein the method is employed in an advertising sales method.

10. One or more non-transitory, tangible computer storage media storing computer-executable instructions for performing a method for determining user purchasing intent, the method comprising:

obtaining a training web page having a first purchasing intention label to train a purchasing intent model, the training web page being a web page publicly accessible via the Internet,

wherein the first purchasing intention label identifies a pre-determined purchasing intent for the training web page,

wherein purchasing intent is a degree of interest, while viewing a web page or conducting a query, that a user has in purchasing one or more products or services related to the web page or query, and

wherein purchasing intent is classified as one of;

purchase, informational, or non-purchase, purchase indicating the user has a strong interest in purchasing a product related to the web page or query, informational indicating the user has moderate interest in purchasing and/or is researching, and non-purchase indicating the user has low or no interest in purchasing;

obtaining a training query having a second purchasing intention label to train the purchasing intent model, the training query being an Internet search query,

wherein the second purchasing intention label identifies a pre-determined purchasing intent for the training query;

extracting one or more purchasing intent indicators training web page, purchasing intent indicators comprising web page content, keyword patterns, web page feature positions, links, and web page layout;

processing the training query by entering the training query into a search engine that returns a search results web page including a list of a top N query result links, each of the top N query result links having an associated Internet web page, where N is an integer from one to infinity;

extracting one or more purchasing intent indicators from the search results web page and each of the Internet web pages associated with the top N query result links;

training the purchasing intent model by applying a machine learning algorithm based on the purchasing intent indicators extracted from the training web page along with the pre-determined purchasing intent identified in the first purchasing intention label as well as the purchasing intent indicators extracted from the search results web page and extracted from each of the Internet web pages associated with the top N query result links along with the pre-determined purchasing intent identified in the second purchasing intention label,

wherein combine coefficients for combining the purchasing intent of the search results web page and each of the Internet web pages associated with the top N query result links are determined such that a combined purchasing intent corresponds to the second known purchasing intent;

constructing a purchasing intent classifier based on the trained purchasing intent model;

obtaining user Internet web page browsing information from a network for purchasing intent classification;

extracting at least one purchasing intent indicator from the user Internet web page browsing information;

using the purchasing intent classifier, determining a purchasing intent for the user Internet web page browsing information based on the at least one extracted purchasing intent indicator;

obtaining user Internet search query information for purchasing intent classification;

processing the user Internet search query information by entering the search query into a search engine that returns a search results web page including a list of a top P query result links, each of the top P query result links having an associated Internet web page, where P is an integer from one to infinity;

extracting one or more purchasing intent indicators from the search results web page and from each of the Internet web pages associated with the top P query results;

determining a purchasing intent for the search results web page and each of the Internet web pages associated with the top P query results using the intent classifier;

determining a purchasing intent for the user Internet search query by combining the purchasing intent of the search results web page and the purchasing intent of each of the Internet web pages associated with the top P query results using the combine coefficients determined by the purchasing intent training model; and

determining a complete user purchasing intent based on the determined purchasing intent of the user Internet web page browsing information and the determined purchasing intent of the user Internet search query information.