Stop Guessing Query Terms and Start Predicting User’s Needs
Usually, when someone uses a search engine, they enter search query terms or a question into a search box. A Patent granted to Google suggests alternatives to “guessing search terms” to try to obtain “information necessary to achieve a particular goal.”
This patent came to my attention as one I wanted to write about because its title included the terms “user’s needs” and “Context”, which I’ve been seeing in several Google patents recently. One of those was about Context Vectors, which Google told us about involved the use of context terms from knowledge bases, to help identify the meaning of terms that might have more than one meaning. An example is Java, which could be a country in Indonesia, a programming language, or a name for coffee. Adding the right context terms on a page about Java could mean Google could more easily identify which meaning of Java a page might be about.
Context is the Search Term of the Year
I’ve been seeing “context” mentioned in some interesting patents so far this year, including for things that I have been watching for patents about. Another post I wrote about Context, was one which told us about Context Facts, in the post My Fifth Post About Context at Google: Adding Context Facts to Question Answers.
In that post, Google showed us that sometimes information that people asked questions about sometimes could supplement their answers with additional information that helped put those facts into context, such as telling us when asked how tall Barack Obama was; instead of just telling us that he was 6’1″, that he was also the 9th tallest president. The post showed us that sometimes carousels that showed contextual facts such as the tallest presidents were things that searchers wanted to see, and Google is interested in predicting user’s needs.
Predicting User’s Needs by Showing ‘People also search for’ in Knowledge Panels
This recently granted patent about predicting a Users’ needs to be based upon a particular context tells us about other things that we might see. For instance, if you search for me. “Bill Slawski”, you will see a knowledge panel which tells you that “people also search for:” and then provides a list of other people who are searched for possibly during the same query sessions that people might search for me, and those people also include Steve Jobs, George Orwell, Aldous Huxley, Gene Roddenberry, Matt Cutts, Tim Berners-Lee, Natzir Turrado, Andrew Isidoro, and Malte Landwehr.
It’s possible that context can show us things we might not have been expecting or anticipating, but which could be helpful.
Why does Google try to anticipate a “User’s Need for a Particular Context?” Google tells us this in the start to the description of the patent:
When a user is in an unfamiliar situation, the user may turn to a computing device for obtaining information and facts that might assist the user in accomplishing a certain task to achieve a particular goal. Some computing devices require that the user be able to provide sufficient information (e.g., search query terms) for guiding the computing device in locating the particular information that the user is searching for.
Unfortunately, the user may be unaware of the tasks the user may need to complete, much less the information for which the user should search, to successfully navigate through the unfamiliar situation and achieve the particular goal. Consequently, without prior knowledge of various actions that the user may need to perform, the user may experience stress and waste valuable time and resources inputting information into a computing device and guessing search terms, as the user tries to obtain information necessary to achieve a particular goal.
In addition to telling us what other people might be likely to search for after performing a search for one thing, We may experience a news feed like the one from Google Now that shows us the latest stories that match interests of ours that may be identified from our previous searches, or interests that we have indicated that we have. These could be work-related, or entertainment-related interests. It’s a different use of fulfilling informational or situational needs of a searcher but may enrich their use of search, by predicting user’s needs.
The patent is:
Predicting user needs for a particular context
Inventors: Yew Jin Lim, James Kunz, Joseph Garrett Linn, Charles Jordan Gilliland, David Faden, Sanjit Jhala
Assignee: GOOGLE LLC
US Patent: 9,940,362
Granted: April 10, 2018
Filed: May 26, 2015
A computing system is described that identifies, based on search histories associated with a group of computing devices for a particular context, a task performed by users of the group of computing devices for a particular context.
The computing system determines the first degree of likelihood of the task being performed by the users of the group of computing devices for the particular context and determines the second degree of likelihood of the task being performed by the users of the group of computing devices for a broader context that includes the particular context and at least one other context.
Responsive to determining that the first degree of likelihood exceeds the second degree of likelihood by a threshold amount and that a current context of a particular computing device corresponds to the particular context, the computing system transmits, to the particular computing device, information for completing the task for the particular context.
Citations from the Predicting User’s Needs Patent
There is a list of articles cited as references from the patent applicants. Based upon their titles and subjects, I decided that it might be worth providing links to those to enable readers to read articles that the patent’s inventors thought were important enough to cite in the patent. Some of those papers are extremely interesting, focusing upon things such as query reformation, recommendation systems, predictive search results, and how intelligent assistants such as Siri, Google Now, and Viv function.
White et al., “Predicting Short-Term Interests Using Activity-Based Search Context,” CIKM’10, ACM, Oct. 23-30, 2010, 10 pp. cited by applicant .
Hong et al., “Context-aware system for proactive personalized service based on context history,” Expert Systems with Applications, vol. 36, No. 4, Dec. 25, 2008, 10 pp. cited by the applicant.
Lane et al., “Hapori: Context-based Local Search for Mobile Phones using Community Behavioral Modeling and Similarity,” Ubicomp 2010, ACM, Sep. 26-29, 2010, 10 pp. cited by the applicant.
Olivarez-Giles, “Google Now` Will Suck in Outside App Data,” The Wall Street Journal, Jan. 30, 2015, 3 pp. cited by the applicant.
Belkin, “Helping People Find What They Don’t Know,” Communications of the ACM vol. 43, No. 8, Aug. 2000, pp. 58-61. cited by the applicant.
Warshaw, “A New Model for Predicting Behavioral Intentions: An Alternative to Fishbein,” Journal of Marketing Research, vol. XVII, May 1980, pp. 153-172. cited by the applicant.
Adomavicius et al., “Context-Aware Recommender Systems,” Proceedings of the 2008 ACM Conference on Recommender Systems, Oct. 23-25, 2008, Switzerland, pp. 1-37. cited by the applicant.
Radwan, “Predicting behavior from actions in the past,” 2KnowMySelf, 2014, Retrieved from: <http://www.2knowmyself.com/predicting_behavior_from_actions_in_the_pa- st> 3 pp. cited by applicant .
Belkin, “Intelligent Information Retrieval: Whose Intelligence?” Proceedings of the Fifth International Symposium for Information (ISI’96), 1996, 6 pp. cited by the applicant.
Levy, “Siri’s Inventors are Building a Radical New AI That Does Anything You Ask,” Wired, Retrieved from: <http://www.wired.com/2014/08/viv/> Aug. 12, 2014, 20 pp. cited by the applicant.
Viv Labs, “Viv: The Global Brain,” Viv Labs, Retrieved from: <http://viv.ai/> 2015, 4 pp. cited by applicant .
White et al., “Predicting User Interests from Contextual Information,” Proceedings of the 32nd International ACM SIGIR conference on Research and Development in Information Retrieval, Jul. 19-23, 2009, Boston, MA, 8 pp. cited by applicant.
Belkin et al., “Relevance Feedback versus Local Context Analysis as Term Suggestion evices: Rutgers’ TREC-8 Interactive Track Experience” Proceedings of 8th Text Retrieval Conference, Washington D.C.; NIST, pp. 1-9, Retrieved on Apr. 15, 2015. cited by the applicant.
Many features that Google is showing context for, around predicting user’s needs, appear to be related to user search history data. It is one of the foundations upon which Siri, Google Now, and Viv function. It’s worth studying how context is used by those applications and in things such as recommendation systems in eCommerce. If Google is successful in predicting user’s needs, it can anticipate searches that people may want to make and leave them satisfied.
Search News Straight To Your Inbox
Join thousands of SEOs to get the best search news in under 5 minutes. Get resources, tips and more with The Splash newsletter: