Author: Olaf Kopp
Reading time: 4 Minutes

Predicting intent of a search for a particular context

Topics: , ,

Rate this post

The patent focuses on a method for predicting the intent of a user’s search query within a specific context and subsequently adjusting search results to emphasize information that satisfies this inferred intent. This is particularly relevant for enhancing the user experience in information retrieval systems by making the search results more relevant to the user’s actual needs or intentions.

  • Patent ID: US10909124B2
  • Country: United States, China, WIPO
  • Date of Patent: February 2, 2021
  • Inventors: Yew Jin Lim, Joseph Linn, Yuling Liang, Carsten Steinebach, Wei Lwun Lu, Dong Hyun Kim, James Kunz, Lauren Koepnick, Min Yang
  • Assignee: GOOGLE LLC, Mountain View, CA (US)
  • Filed: May 18, 2017


The background section of the patent discusses the challenges users face when seeking information using a computing device. Users often rely on these devices to assist them in accomplishing tasks by providing relevant information and facts. However, the effectiveness of the search depends significantly on the user’s ability to provide sufficient and precise information through search queries. If the search query is too broad or not well-defined, the computing device may return an overwhelming amount of information, making it difficult for users to find the most relevant or interesting data. This situation can lead to frustration, as users may have to input detailed queries, execute multiple searches, or sift through large quantities of search results to find the necessary information to complete their tasks. The patent aims to address these issues by improving how search intents are predicted and how search results are adjusted to meet user needs more effectively.


  • Intent Prediction: The method involves determining the intent of a search query using a computing system that analyzes user-initiated actions performed across multiple devices. This determination is made based on contextual information and the specific search query received from a computing device.

The patent considers a wide array of contextual information to predict the intent of a search query accurately and adjust the search results accordingly. This contextual information includes but is not limited to:

    • Location Data: Information about the user’s current location or locations associated with the search query. This can help determine if the search intent is local (e.g., looking for nearby services or information specific to a geographic area).
    • Time of Day: The time when the search is conducted, which can influence the relevance of certain search results (e.g., searching for coffee shops early in the morning might imply a preference for those currently open).
    • Device Information: The type of device used for the search (e.g., mobile phone, laptop), which might affect the user’s intent (e.g., mobile searches might have more local and immediate needs).
    • Operating State: Information about the device’s current operating state, such as battery level, connectivity status, or whether specific apps are running, which could influence the context of a search.
    • User Behavior: Past user behavior and interactions with the device, including application usage patterns, search history, and other user-initiated actions that can provide insights into current search intentions.
    • Environmental Conditions: External conditions such as weather or traffic status, which might be relevant for searches related to outdoor activities or commuting.
    • Calendar and Time-Specific Events: Information from the user’s calendar or notable dates that might influence search intent (e.g., searching for “flowers” close to Valentine’s Day might imply an intent to purchase).
    • Sensor Data: Inputs from device sensors, such as an accelerometer, gyroscope, or light sensor, which could indicate the user’s current activity level or environment, further refining the context of a search.
    • Social and Communication Context: Information derived from social media activity or communication patterns, such as recent posts or messages, which could suggest current interests or concerns.
  • Adjusting Search Results: Once the intent is determined, the method includes adjusting at least a portion of the search results obtained from the query. The adjustment emphasizes information that satisfies the identified intent, making the search results more relevant and useful to the user.
  • Sending Adjusted Results: The adjusted search results are then sent back to the user’s computing device, providing an indication of the tailored results. This ensures that the user receives information that is more likely to be relevant to their query and intended task.
  • Machine Learning Models: The claims also touch upon the use of machine-learned models, including deep learning models, trained on log data indicative of user inputs received by a group of computing devices. This training helps to accurately predict the user’s intent based on past user actions and contextual information.
  • Data Inclusion: The patent mentions the inclusion of various types of data in the process, such as application usage data and search feature data for different contexts. This data contributes to the system’s ability to learn from and predict user behavior more effectively.
  • Implementation: The system described includes at least one processor and a memory containing instructions, which, when executed, enable the prediction of search intent and the adjustment of search results accordingly.

Implications for SEO

  • Understanding User Intent: SEO strategies must evolve to focus more on understanding and addressing specific user intents rather than merely targeting keywords. Content should be crafted to answer questions, solve problems, or meet needs that users might have in various contexts.
  • Content Diversification: To cater to different user intents, diversify content types (e.g., informational blogs, tutorials, product pages) to cover a broader range of search scenarios and intents.
  • Personalized Content Strategy: As the system uses past user behavior to predict search intent, creating personalized content experiences becomes crucial. SEO strategies could involve developing user personas and tailoring content to match the interests and behaviors of target segments.
  • Engagement and Interaction: Encouraging user engagement and interaction with content can provide valuable behavioral data that could influence how search engines interpret the intent behind future searches. Implementing interactive elements, user feedback options, or social sharing capabilities could be beneficial.


Content from the blog

LLMO: How do you optimize for the answers of generative AI systems?

As more and more people prefer to ask ChatGPT rather than Google when searching for read more

What is the Google Knowledge Vault? How it works?

The Google Knowledge Vault was a project by Google that aimed to create an extensive read more

What is BM25?

BM25 is a popular ranking function used in information retrieval systems to estimate the relevance read more

The dimensions of the Google ranking

The ranking factors at Google have become more and more multidimensional and diverse over the read more

Interesting Google patents for search and SEO in 2024

In this article I would like to contribute to archiving well-founded knowledge from Google patents read more

What is the Google Shopping Graph and how does it work?

The Google Shopping Graph is an advanced, dynamic data structure developed by Google to enhance read more