Voice search and digital assistants like Google Assistant are becoming increasingly significant, influencing how search queries are conducted across various devices, including cars, home appliances, and Apple devices. This shift towards voice search raises an essential question – is your content ready for this new environment?
In December 2017, Google released its Human Raters Guidelines, which evaluate the algorithm producing Google Assistant search results. When users perform a voice search, the algorithm often summarizes the result in a spoken format.
This announcement referenced a research paper titled, "Sentence Compression by Deletion with LSTMs," detailing the algorithm behind these voice search summaries. The "compression system" involved removes words and phrases to create useful summaries.
The research paper outlines the types of content that cannot be efficiently summarized and specifies what kinds of words and phrases are removed. Publishers looking to make their content more accessible for voice search may find this information particularly useful, given the growing importance of voice search.
Key Elements of Voice Search Summaries
Google Assistant’s voice search summaries are composed of four essential elements:
- Length: The summary should be concise and appropriate for spoken format.
- User Satisfaction: The information provided must meet the user’s needs.
- Grammatical Correctness: The sentences should be well-formed and coherent.
- Elocution: The Google Assistant software should pronounce the summary correctly.
How Voice Search Summarizes Content
The research paper noted that the summarizing algorithm doesn’t rely on explicit syntax features but translates words into machine-readable tokens. It then removes unnecessary words and phrases to create a summary – a process called Compression. Unlike standard syntactic methods, this system focuses on token deletion decisions.
The compression model’s efficiency is supported by high readability and informativeness scores assigned by human raters, even though it doesn’t use explicit syntactic information. The research also found that integrating grammatical features did not necessarily improve the algorithm’s performance.
Example of Content Summary
Here’s a practical demonstration:
- Original Content: "Alan Turing, known as the father of computer science, the codebreaker that helped win World War 2, and the man tortured by the state for being gay, is to receive a pardon nearly 60 years after his death."
- Summarized Content: "Alan Turing is to receive a pardon."
Challenges in Summarizing Content
Certain types of content pose difficulties for the summarizing algorithm, such as those with quotes, commas, or complex punctuation. For instance:
- Original Content: "A Virginia state senator and one-time candidate for governor stabbed by his son said Friday that he is ‘alive so must live,’ his first public statement since the assault and his son’s suicide shortly thereafter."
- Summarized Content: "State senator alive so must live."
Content featuring too many commas also challenges the algorithm. The complexity typically arises from the frequency of punctuation rather than the punctuation itself.
Factors Affecting Voice Search Summaries
The research indicated four primary features that complicate content summarization:
- Commas
- Quotes
- Lack of removable content
- Complex event context
An example illustrating this complexity:
- Original Content: "Just a week after a CISF trooper foiled a suicide bid by a woman in the Delhi metro, another woman trooper from the same force prevented two women commuters from ending their lives, an official said Monday."
- Summarized Content: "Another woman trooper prevented two women commuters."
Types of Words and Phrases Commonly Removed
- Appositions: Phrases that have a direct relationship, such as "a very popular singer" in "Dean Martin, a very popular singer, will be performing at the Sands Hotel."
- Temporal Expressions: Phrases indicating time, such as dates or hours.
- Optional Modifiers: Generally adjectives (modifying nouns) and adverbs (modifying verbs).
- Introductory Clauses: Phrases setting up a statement, like "If they want to win, athletes must exercise every day."
Optimizing Content for Voice Search Summaries
There’s no definitive formula for writing content tailored for Google Assistant. However, avoiding overly long or complex sentences and ensuring readability can be beneficial. Utilizing writing and grammar tools might help, especially if you don’t have access to professional editing. Ensuring your content sounds natural when read aloud can also enhance its suitability for voice search.
Final Notes
Writing content that balances readability and succinctness can help in making it more adaptable for voice search. Monitoring grammar and structure is beneficial, as these factors play a crucial role in how efficiently content is summarized for voice search results.
Images by Shutterstock, modified by Author.
