MasterObjects, Inc. v. Amazon.com, Inc.
- William Alsup
- 3:20-cv-08103
- U.S. District Court · Northern District of California
- 17
In MasterObjects v. Amazon, Judge Alsup granted Amazon’s summary-judgment motion on literal patent infringement, while leaving equivalent-infringement issues open.
MasterObjects lost its literal-infringement claims at summary judgment because the court found Amazon’s autocomplete system lacked the required cache and underlying data source. Amazon obtained summary judgment on literal infringement; the order did not decide infringement under the doctrine of equivalents.
What happened
MasterObjects, Inc. accused Amazon.com, Inc. of infringing three patents covering autocomplete technology. Amazon asked the court to rule without a trial that its system did not infringe.
The court interpreted the patents’ use of “cache” to require storing both prior autocomplete queries and their results. It found that Amazon’s system used databases containing generated prefixes and product-search data, not stored prior autocomplete queries and results. The court also found that Amazon lacked the separately queryable underlying data source required by the patent claims.
Judge Alsup granted Amazon’s summary-judgment motion on literal infringement of all asserted claims. The motion did not address infringement under the doctrine of equivalents, so the court ordered the parties to submit statements about how to proceed on that issue and left the trial schedule in place for the time being.
The detailed version
- MasterObjects, Inc. v. Amazon.com, Inc. · No. 3:20-cv-08103
- William Alsup
- Sept. 15, 2022
Background
MasterObjects, Inc. sued Amazon.com, Inc. for allegedly infringing three patents involving autocomplete technology: U.S. Patent Nos. 9,760,628; 10,311,073; and 10,394,866. The technology suggests completed search queries as a user types. Amazon moved for summary judgment of noninfringement, which asks the court to decide that no reasonable trial could establish infringement on the record presented.
The dispute centered on the patents’ “cache” limitations. The court described a cache as a memory store that saves a copy of both an autocomplete query and the results generated from an underlying data source. The court also concluded that the claimed system requires an underlying data source that can itself be queried for autocomplete results.
The parties’ systems
Amazon’s autocomplete system uses paired read-only databases created during a process Amazon calls the “daily build.” One database, the Prefix RODB, contains generated prefixes linked to entries in the Keyword RODB, which contains autocomplete suggestions. The daily build creates prefixes through progressive letter construction. For example, a keyword such as “Rolex” produces “R,” “Ro,” “Rol,” “Role,” and “Rolex.”
The databases are built from Tommy Query Groups, which contain completed product-search strings and related metadata. The court found that Tommy does not store autocomplete queries and, at most, stores autocomplete results. Amazon also uses Personalization Platform Data Service data to re-rank suggestions based on a user’s prior completed product searches.
Claim construction and noninfringement
The court construed “cache” in context to mean a particular memory store that saves a copy of both the autocomplete query and the results from the data source. It relied on the language of the asserted claims, the patents’ common specification, and statements previously made by MasterObjects and its expert. The court noted that MasterObjects had earlier described the claimed cache as storing previous queries and results, but argued in opposition to summary judgment that only prior autocomplete results were required.
The court rejected that narrower position. It reasoned that claim language referring to looking up a query in the cache necessarily indicated that the prior query had also been stored. It also found that the generated prefixes in Amazon’s Prefix RODB were not copies or a history of prior autocomplete queries. They were artificially generated prefixes that could lead to a keyword.
The court separately found that Amazon’s system did not use the required underlying data source. Tommy Query Groups could not itself be queried for autocomplete results, and neither the Personalization Platform Data Service nor the raw Tommy data could be queried for those results. The court rejected MasterObjects’ argument that a cache could also serve as its own content source because, under the claims, something must underlie the cache and be independently queryable.
The court further concluded that MasterObjects could not prove infringement at trial under the adopted claim construction. Its expert had not provided an infringement opinion under Amazon’s proposed construction and had not identified a qualifying underlying data source. MasterObjects had stated that it would rely only on its retained experts to prove infringement. The court therefore found no relevant expert testimony supporting infringement under its construction.
Disposition
The court granted Amazon’s motion for summary judgment of noninfringement. The order stated that this disposed of literal infringement of all claims-in-suit. It also stated that Amazon’s motion did not address infringement under the doctrine of equivalents. Under Rule 56(f), the parties were ordered to submit statements within five calendar days concerning how to handle that issue. The court said the trial and pretrial dates remained in place for the time being and that it might later vacate the trial date.
Read the full 17-page opinion on CourtListener, the free public archive maintained by the Free Law Project.