Google Challenges: Understanding not and prepositional queries

In her recent Google's Search Off The Record podcast, Elizabeth Tucker, director and product manager at Google, noted that despite the company's best efforts to match users' queries with the information they're looking for, some challenges still remain. Requests containing the word "not" and prepositions are particularly difficult to understand, Tucker says.

"Prepositions, in general, are another hard one. And one of the really big, exciting breakthroughs was the BERT paper and transformer-based machine learning models when we started to be able to get some of these complicated linguistic issues right in searches."

BERT (Bidirectional Encoder Representations from Transformers) is a natural language processing technique based on a neural network that Google started using in search in 2019. This technology is designed to understand the nuances and context of words in search queries, rather than to interpret queries as a collection of individual terms. 🚀

• 📌 Regardless of the potential of BERT and similar advances, Tucker admits that Google's ability to analyze complex queries is still a work in progress. In particular, queries with the word "not" continue to cause problems for the search engine.

For example, Google algorithms can interpret a search query like "shoes not made in China" in different ways. Does the user want shoes made in countries other than China, or is he looking for information on why some shoe brands have moved their production out of China? This ambiguity creates a challenge for websites trying to rank for such queries. 🚀

• 📌 Another area where Google's algorithms can get it wrong is prepositions, which show the relationship between words in a sentence. Queries like "restaurants with outdoor terraces" or "hotels near the beach" rely on prepositions to convey key information about the user's needs.

For SEO professionals, this means that optimizing for prepositional queries may require additional effort. It is not enough to simply include relevant keywords on a page; the content should be structured in such a way as to convey the relationship between these keywords. 🚀

• 📌 The importance of Google's understanding of complex queries is especially great for long search queries - those highly specific, often wordy phrases that make up a significant portion of all search traffic. Long-tail keywords are often seen as a golden opportunity for SEO because they have less competition and can signal a high level of user intent.

However, if Google can't understand these complex queries, it can be harder for websites to rank for them, even with well-optimized content. 🚀

"I would not say this is a solved problem. We're still working on it."

Tucker notes that Google is actively working to improve the processing of these linguistically complex queries, but a complete solution may still be a long way off. Meanwhile, users may have to rephrase their search queries or try different query phrases to find the information they're looking for—frustrating in an age when many expect Google to intuitively understand their needs.