MR

Snippets in Search Engine Results

  • Snippets: Short descriptions/excerpts from websites in search engine results.

    • Introduced by Google in 1998.

    • Automatically generated based on site content and query terms.

    • Key query terms are bolded.

    • Computed at query time; vary by query.

    • Content from anywhere on the page.

    • Forums: May display post count, authors, date.

    • Scholarly articles: May display author, year, citation.

  • Snippets as Summarization

    • Automatic summarization.

    • Document summarization.

    • Two approaches: Extraction: Selecting existing text, Abstraction: Building a semantic representation.

    • Primarily focuses on extractive methods.

  • Featured Snippets

    • Answers query directly on the search results page.

    • Introduced in 2016.

    • Appears above #1 spot.

    • Pulled from Page 1 results.

    • Not necessarily the #1 result.

    • Excerpt best answers the query.

    • Three Types:

    • Paragraph: Direct definition.

    • List: Bulleted/numbered list.

    • Table: Data in table format.

  • Modifying Your Page to Produce a Featured Snippet

    • Simple on-page adjustments.

    • Fuels voice search.

    • Tips:

    1. Add "What Is [Keyword]" heading.

    2. Use the "is statement".

    3. Define the topic in 2-3 sentences.

    4. Match the format (paragraph, list, table).

    5. Don't use first person.

  • TLDR (Too Long Didn't Read)

    • For popular queries:

    • Paid ads

    • Featured snippet

    • People Also Ask

    • Portion of #1 organic result

    • Difficult to be found in organic search.

    • Above the fold: visible without scrolling.

  • Extracting a Snippet is Not Always Easy

    • Example: "Tesla reports financial results" using "earnings".

    • Example: "cloud computing" snippet from meta-description.

  • How Does Google Generate Snippets?

    • Search for Google patents with the term "snippets".

  • Guidelines for Snippet Generation (Based on US Patent 8,145,617)

    • Location-Based Rules:

    • Based on the location of the query terms in the page.

    • Paragraph scoring.

    • Documents with abstracts, summaries, introductions.

    • Ends of pages with conclusion or summarization.

    • Language-Dependent Rules:

    • Punctuation, sentence endings.

    • Word length.

    • Bold or italicized words.

    • Rejection Rules:

    • Short snippets.

    • Mostly punctuation.

    • Too many bold/italicized words.

  • Algorithm for Snippet Generation (US Patent 8,145,617)

    1. Identify paragraphs with query terms.

    2. Score paragraphs.

    3. Return phrase with the highest score.

    • Selects paragraph near beginning if there is abstract, executive summary, or long introduction.

    • Uses end of document when there is a conclusion or summarization at the end.

    • Scoring includes:

    • Paragraphs shorter than a threshold score 0.

    • The k-th paragraph from the start gets a score of kth{\text -position Factor + max(actual paragraph length, maxParagraphLen)}

    • The paragraph with the highest score is selected for the snippet.

  • Snippets Based on User Profile (US Patent 8,631,006)

    • Personalized based on user's profile.

    • Terms related to user's profile are identified.

    • Term profile compared to user's profile.

    • Similarity score.

    • High-scoring terms used in identifying snippets.

    • Terms may be added to search terms.

  • Featured Snippet Study

    • Paragraph snippets are the most common.

    • Position 1 is the most common ranking position.

    • Image thumbs are most frequent other snippets.

  • Google's People Also Ask (PAA) Feature

    • Introduced in 2015.

    • PAA Result for query "how does google generate snippets".

    • The "People Also Ask" box appeared on 364 keywords out of 1,788 (20%).

  • People Also Ask (PAA) is Growing Fast

    • Google universal SERP result.

    • Related to the featured snippet

    • Each box contains one to four related questions.

    • Site URL appears below each answer.

    • PAAs are growing faster than snippets.

  • Rich Snippets

    • Announced in 2009.

    • Allows developers to include information for Google's algorithm.

    • Provides a summary of information.

    • Rich Snippets Examples: People Snippets

    • Snippet describes the pages with information about the individual, such as Facebook, LinkedIn, Google.

    • Rich Snippets Examples: Events

    • The Fillmore theater can highlight future concerts by regularly updating their webpage with the latest rich snippet information.

  • Advantages of Rich Snippets

    • Webmasters: Add useful information.

    • Purpose: Provide more information to users.

    • Two good reasons for using rich snippets:

    1. Additional traffic to a webpage

    2. Higher Click Through Rate

  • Joint Effort by Google, Yahoo! and Bing

    • Agreed on a single standard in June 2011.

    • Established schema.org.

    • Standardized on the microdata format.

  • Rich Snippet Technology Definitions

    • Google suggests using the microdata formalism for snippets.

    • Two other formalisms:

    • RDFa

    • Microformat Encoding

  • Schema.org Vocabulary

    • Defines an object hierarchy.

    • Most general item: Thing (name, description, url, image).

    • Person, Place, and Organization are types of Things.

    • More specific items inherit properties.

    • Commonly used types:

    • Creative works: book, movie, music recording, recipe, TV Series

    • Embedded object: image, video

    • Event

    • Organization

    • Person

    • Place, Local Business, e.g., Restaurant

    • Product, Offer, Aggregate Offer

    • Review, AggregateRating

  • Entities in Rich Snippet Encodings

    • Supported by Google:

    • Software applications

    • Breadcrumbs

    • Events

    • Music

    • Businesses and Organizations

    • People

    • Products

    • Recipes

    • Review Ratings

    • Reviews

    • Videos

  • Rich Snippets - Microformats and Microdata

    • Microformats: Use existing HTML (e.g., class attribute).

    • Microdata: Extends HTML5 with new attributes like itemprop.

  • MicroData Example: A Web Page About the Movie Avatar

    • Add itemscope to the HTML tag.

  • Avatar Example Continued

    • Use itemprop to label properties.

  • List of Movie Properties (Schema.org/Movie)

    • Properties: description, image, name, url, etc.

  • MicroData Markup for "Pirates of the Caribbean"

    • Includes: movie name, description, director, author, actors.

  • More Examples: Clarifying Hard to Understand Content

    • The `` element has attributes: dates, times, durations.

  • Schema Markup Generator

    • Web interface tool.

  • Google's Rich Snippets Testing Tool

    • Tools for testing structured data.

  • Structured Data Testing

    • Tool for examining microformat data.

  • Google Introduces New Tags for Snippet Control

    • Robots meta tag:

    • nosnippet: No textual snippet.

    • max-snippet: [number]: Max text length.

    • max-video-preview: [number]: Max video duration.

    • max-image-preview: [setting]: Max image size.

    • data-nosnippet: HTML attribute on span, div, section.

    • nosnippet blocks all snippets for the tagged page.

    • If both nosnippet and data-nosnippet appear in a page, nosnippet takes priority.

  • Summary

    • Five categories:

    1. Regular snippets

    2. Rich snippets

    3. Google News

    4. Entity types

    5. Features snippets

  • Extending Schema.org to handle PAA

    • QAPage

    • Question

    • HowTo