DCT
1:17-cv-00691
Word To Info Inc v. Amazon.com Inc
I. Executive Summary and Procedural Information
- Parties & Counsel:
- Plaintiff: Word to Info, Inc. (Texas)
- Defendant: Amazon.com, Inc. (Delaware)
- Plaintiff’s Counsel: STAMOULIS & WEINBLATT LLC
- Case Identification: 1:17-cv-00691, D. Del., 06/06/2017
- Venue Allegations: Venue is alleged to be proper in the District of Delaware because Amazon is a Delaware corporation that conducts substantial business and has allegedly committed acts of infringement in the district.
- Core Dispute: Plaintiff alleges that Defendant’s Amazon Alexa personal assistant software infringes a family of seven patents related to methods for processing natural language to store and retrieve knowledge.
- Technical Context: The patents-in-suit describe foundational technologies for natural language understanding systems, which are central to the functionality of modern digital voice assistants.
- Key Procedural History: The complaint alleges that at least one of the patents-in-suit was cited as prior art during the patent prosecution of several U.S. patents assigned to Defendant Amazon or companies it later acquired. These allegations may be used to support claims of pre-suit knowledge and willful infringement.
Case Timeline
| Date | Event |
|---|---|
| 1994-09-30 | Earliest Priority Date for all Patents-in-Suit |
| 1998-02-03 | ’468 Patent Issued |
| 2000-10-24 | ’087 Patent Issued |
| 2003-08-19 | ’091 Patent Issued |
| 2005-08-01 | True Knowledge / Evi Founded |
| 2008-03-25 | ’840 Patent Issued |
| 2011-01-18 | ’509 Patent Issued |
| 2012-10-01 | Evi technology acquired by Amazon |
| 2012-12-04 | ’603 Patent Issued |
| 2014-04-01 | ’436 Patent Issued |
| 2017-06-06 | Complaint Filed |
II. Technology and Patent(s)-in-Suit Analysis
U.S. Patent No. 5,715,468 - "Memory System for Storing and Retrieving Experience and Knowledge with Natural Language"
- Issued: February 3, 1998 (’468 Patent)
The Invention Explained
- Problem Addressed: The patent’s background section describes prior art natural language processing systems as being limited in their ability to understand complex language and expand their knowledge base, often because they rely on rigid, predefined structures that cannot effectively represent and combine multiple natural language inputs into a coherent body of experience and knowledge (Compl., Ex. A, '468 Patent, col. 2:37-51).
- The Patented Solution: The invention proposes a system that processes natural language by first converting it into electronically encoded data. This data is then processed using a specialized dictionary and a set of grammar rules to produce a "grammatical parse," which represents the syntactic structure of the language input. This parsed data forms the basis for storing and retrieving information in a more flexible and expandable memory system ('468 Patent, Abstract; col. 4:26-45).
- Technical Importance: The described method aimed to create a more robust and scalable framework for machine understanding of language, moving beyond simple keyword matching toward systems capable of syntactic analysis as a foundation for knowledge storage and retrieval ('468 Patent, col. 3:20-33).
Key Claims at a Glance
- The complaint asserts independent claim 1 (Compl. ¶18).
- The essential elements of independent claim 1 are:
- A method of processing natural language, which comprises steps:
- providing electronically encoded data which is representative of said natural language;
- providing a dictionary data base containing a plurality of entries with syntax usage data, word representation data, and/or function codes;
- lexically processing said electronically encoded data to access said dictionary data base;
- providing a grammar specification; and
- utilizing said syntax usage data from entries of said dictionary data base with reference to said grammar specification to produce output data representing a grammatical parse of said natural language.
- The complaint also asserts dependent claims 8, 21, 29, and 33 and reserves the right to assert additional claims (Compl. ¶18).
U.S. Patent No. 6,138,087 - "Memory System for Storing and Retrieving Experience and Knowledge with Natural Language Utilizing State Representation Data, Word Sense Numbers, Function Codes and/or Directed Graphs"
- Issued: October 24, 2000 (’087 Patent)
The Invention Explained
- Problem Addressed: As a continuation of the application leading to the ’468 Patent, this patent addresses the same challenge of enabling computers to understand natural language in a comprehensive, rather than limited, fashion ('087 Patent, col. 2:25-46).
- The Patented Solution: The '087 Patent builds upon the earlier system by further defining the data structures used for comprehension. It specifies a dictionary containing "word sense numbers" which are linked to "state representation data." This allows the system to not just parse language syntactically, but also to access the specific meaning or "state" of a word (e.g., the state of a noun, verb, or adjective) within the context of the input ('087 Patent, Abstract). This knowledge is stored in a memory system that can include directed graphs to represent relationships between concepts ('087 Patent, col. 4:4-24).
- Technical Importance: By introducing the concepts of "word sense numbers" and "state representation data," the invention provided a more granular method for a machine to disambiguate words and understand their specific meaning in context, a critical step for more advanced language processing ('087 Patent, col. 4:50-62).
Key Claims at a Glance
- The complaint asserts independent claim 17 (Compl. ¶26).
- The essential elements of independent claim 17 are:
- A method of processing natural language, which comprises steps:
- providing electronically encoded data which is representative of said natural language;
- providing a dictionary data base containing entries with syntax usage data and word sense numbers having associated state representation data;
- lexically processing said electronically encoded data to access said dictionary data base; and
- utilizing said syntax usage data and said word sense numbers from said dictionary to access said state representation data.
- The complaint also asserts dependent claim 18 (Compl. ¶26).
Multi-Patent Capsule Analysis
Patent Identification: U.S. Patent No. 6,609,091, "Memory System for Storing and Retrieving Experience and Knowledge with Natural Language Utilizing State Representation Data, Word Sense Numbers, Function Codes and/or Directed Graphs," issued August 19, 2003 (’091 Patent).
- Technology Synopsis: The ’091 Patent, a continuation in the same family, further develops the natural language processing system. It introduces a method where word sense numbers are selected only if their associated state representation data meets certain "requirements," thereby providing a more sophisticated mechanism for disambiguating language based on predefined constraints or rules (Compl. ¶40).
- Asserted Claims: Claims 1 and 12 (Compl. ¶34).
- Accused Features: The complaint alleges that Alexa infringes by providing a "database of requirements" that must be met for certain entities (word sense numbers) to be selected based on user queries and syntactic analysis (Compl. ¶40).
Patent Identification: U.S. Patent No. 7,349,840, "Memory System for Storing and Retrieving Experience and Knowledge with Natural Language Utilizing State Representation Data, Word Sense Numbers, Function Codes, Directed Graphs and/or Context Memory," issued March 25, 2008 (’840 Patent).
- Technology Synopsis: The ’840 Patent adds the concept of a "context database" to the system. This allows the processor to utilize conversational context—such as user location, time, or the subject of a previous query—to more accurately select the correct word sense numbers and understand the user's intent (Compl. ¶51).
- Asserted Claims: Claims 1, 2, 3, and 5 (Compl. ¶44).
- Accused Features: Alexa is accused of infringing by providing a "context database" that uses query contexts, such as a request for "the weather today" or a user's location, to present context-specific data to the user (Compl. ¶51).
Patent Identification: U.S. Patent No. 7,873,509, "Memory System for Storing and Retrieving Experience and Knowledge with Natural Language Utilizing State Representation Data, Word Sense Numbers, Function Codes, Directed Graphs, Context Memory, and/or Purpose Relations," issued January 18, 2011 (’509 Patent).
- Technology Synopsis: The ’509 Patent focuses on the architecture of the knowledge database, describing it as comprising "directed graphs with nodes" that have "access conditions." The system performs "relation path identification processing" to find paths through the graph, where the access conditions on the nodes must be met to proceed, thereby retrieving structured information (Compl. ¶¶ 56, 58).
- Asserted Claims: Claims 9, 10, and 16 (Compl. ¶55).
- Accused Features: Alexa is alleged to infringe by utilizing directed graphs (i.e., a knowledge graph) with paths and access conditions, and by performing path identification to find and select information based on criteria, such as selecting a restaurant entity based on a "Pizzeria" genreId (Compl. ¶¶ 56, 60).
Patent Identification: U.S. Patent No. 8,326,603, "Memory System for Storing and Retrieving Experience and Knowledge with Natural Language Queries," issued December 4, 2012 (’603 Patent).
- Technology Synopsis: The ’603 Patent describes the interaction between a natural language processor and the directed graph database. The processor provides natural language associated with "a clause implying word sense numbers," which is then used to identify and traverse paths within the knowledge graph to fulfill the query (Compl. ¶¶ 64-66).
- Asserted Claims: Claims 14 and 16 (Compl. ¶63).
- Accused Features: Alexa's natural language processor is accused of providing language with associated word sense numbers and using them to identify paths in its knowledge database where access conditions are met (Compl. ¶¶ 64, 66).
Patent Identification: U.S. Patent No. 8,688,436, "Memory System for Storing and Retrieving Experience and Knowledge by Utilizing Natural Language Responses," issued April 1, 2014 (’436 Patent).
- Technology Synopsis: The ’436 Patent introduces a "natural language plausibility and expectedness processor." This component is used to initiate access to the dictionary database by evaluating the likelihood of a given input, which can be used to disambiguate words or handle incorrectly spelled queries based on context (Compl. ¶74).
- Asserted Claims: Claims 1, 2, and 7 (Compl. ¶69).
- Accused Features: Alexa is alleged to infringe by providing a plausibility and expectedness processor that can, for example, answer incorrectly spelled queries or disambiguate words like "today" and "my" based on the context of the user's request (Compl. ¶74).
III. The Accused Instrumentality
Product Identification
- Defendant's "Amazon Alexa personal assistant software" (Compl. ¶18).
Functionality and Market Context
- The complaint alleges that Amazon Alexa is a software system that receives natural language input, such as speech, and converts it into electronically encoded data for processing (Compl. ¶19). The system is alleged to utilize a "dictionary database" based on Evi technology, which was founded as "True Knowledge" and acquired by Amazon (Compl. ¶20). This database is described as containing a plurality of entities and relationships organized in a graph database structure (Compl. ¶¶ 20, 28). The complaint alleges that Alexa lexically and syntactically processes the encoded language input with reference to a grammar specification to produce a grammatical parse, understand user intent, and provide a response (Compl. ¶¶ 21-23). Alexa is positioned in the market as a personal assistant, a central component of Amazon's Echo devices and other consumer electronics.
IV. Analysis of Infringement Allegations
No probative visual evidence provided in complaint.
'468 Patent Infringement Allegations
| Claim Element (from Independent Claim 1) | Alleged Infringing Functionality | Complaint Citation | Patent Citation |
|---|---|---|---|
| providing electronically encoded data which is representative of said natural language | Amazon Alexa encodes natural language speech inputs into audio files and/or text files, such as 16bit Linear PCM audio and HTTP/2 multipart encoded messages. | ¶19 | col. 26:50-67 |
| providing a dictionary data base wherein said dictionary data base contains a plurality of entries which are comprised of one or more of syntax usage data, associated word representation data and/or function codes | Amazon Alexa provides a dictionary database based on Evi and True Knowledge technology, which contains entries having syntax usage data based on entities and their relationships. | ¶20 | col. 27:58-28:9 |
| lexically processing said electronically encoded data to access said dictionary data base | Amazon Alexa lexically processes the encoded data to access its dictionary database. | ¶21 | col. 27:1-13 |
| providing a grammar specification | Amazon Alexa provides a grammar specification for developing skills and defining syntax. | ¶22 | col. 32:20-35 |
| utilizing said syntax usage data from said dictionary data base ... with reference to said grammar specification to produce output data representing a grammatical parse of the natural language | Amazon Alexa utilizes a syntactic parser and syntax usage data from its database, with reference to its grammar, to produce a grammatical parse of speech input. | ¶23 | col. 32:20-46 |
'087 Patent Infringement Allegations
| Claim Element (from Independent Claim 17) | Alleged Infringing Functionality | Complaint Citation | Patent Citation |
|---|---|---|---|
| providing electronically encoded data which is representative of said natural language | Amazon Alexa encodes natural language speech inputs into audio files and/or text files, such as 16bit Linear PCM audio. | ¶27 | col. 26:50-67 |
| providing a dictionary data base wherein said dictionary data base contains a plurality of entries which are comprised of one or more of syntax usage data, associated word sense numbers having associated state representation data, and/or function codes | Amazon Alexa provides a dictionary database (based on Evi/True Knowledge) containing syntax usage data, with entities and relationships that function as word sense numbers associated with state representation data. | ¶28, ¶31 | col. 28:56-66 |
| lexically processing said electronically encoded data to access said dictionary data base | Amazon Alexa lexically processes the encoded data to access its dictionary database. | ¶29 | col. 27:1-13 |
| utilizing said syntax usage data and said word sense numbers which are from entries of said dictionary data base ... to access said state representation data | Amazon Alexa utilizes syntax usage data and word sense numbers from its database to select and access associated state representation data for words in the natural language input. | ¶31 | col. 7:60-65 |
- Identified Points of Contention:
- Scope Questions: A recurring question may be whether the functional descriptions of Alexa's operations, drawn from high-level developer documents and technical papers, map onto the specific, multi-step processes and data structures claimed in the patents. For instance, does Alexa's use of a "graph database" (Compl. ¶28) necessarily implement the "directed graphs comprised of nodes with associated clause implying word sense numbers" as recited in later patents of the family?
- Technical Questions: The patents describe a detailed, integrated architecture with specific memory components (e.g., "Dictionary 20," "Context Memory 120"). A point of contention may be what evidence the complaint provides that Alexa's system, which incorporates acquired technologies like Evi and database platforms like TitanDB, actually implements these specific structures and their claimed inter-operations, as opposed to achieving a similar result through a different architecture.
V. Key Claim Terms for Construction
The Term: "dictionary data base" (from ’468 Patent, Claim 1)
- Context and Importance: This term is fundamental to the earliest asserted claim. The infringement case rests on whether Amazon's knowledge graph, which the complaint alleges is derived from Evi/True Knowledge and built on platforms like DynamoDB and TitanDB (Compl. ¶20, ¶28), falls within the scope of this term. Practitioners may focus on this term because its construction will determine whether a modern, distributed graph database is equivalent to the specific "Dictionary 20" structure disclosed in the patent.
- Intrinsic Evidence for Interpretation:
- Evidence for a Broader Interpretation: The claim language itself is functional, requiring a database that "contains a plurality of entries which are comprised of one or more of syntax usage data, associated word representation data and/or function codes" (’468 Patent, col. 367:1-6). This language could support an interpretation that covers any data structure that performs this role in a natural language system.
- Evidence for a Narrower Interpretation: The specification provides a detailed description of "Dictionary 20," including its specific fields such as "wordsets," "state/function addresses," and "anomalies," and its relationship with other specific system components ('468 Patent, FIG. 3A; col. 27:58-29:21). This detailed disclosure of a specific embodiment may be used to argue for a narrower construction limited to that structure.
The Term: "word sense numbers having associated state representation data" (from ’087 Patent, Claim 17)
- Context and Importance: This concept is central to how the patented system disambiguates language. The dispute will likely focus on whether the method Alexa uses to represent entities and their properties in a knowledge graph is equivalent to the patent's more specific architecture of using numeric "word sense numbers" that serve as pointers to "state representation data" ('087 Patent, Abstract).
- Intrinsic Evidence for Interpretation:
- Evidence for a Broader Interpretation: The patent describes "word sense number" functionally as "an address to the meaning of a word," and "state representation" as data associated with nouns, verbs, and adjectives ('087 Patent, Abstract). This may support a construction that reads on modern object-oriented or entity-relationship models where an entity ID points to a set of attributes.
- Evidence for a Narrower Interpretation: The specification provides highly detailed formats for these data structures, such as the "Concrete Noun Word Sense Number Format" which includes a "Class Number," "Type Number," "Specificity Number," and "Experience Number" ('087 Patent, FIG. 17A; col. 147:1-148:67). This may support an argument that the claims require this specific numeric and categorical structure, which may be architecturally distinct from the accused system.
VI. Other Allegations
- Willful Infringement: The complaint alleges that Amazon’s infringement has been willful and requests enhanced damages (Compl. p. 21, ¶B). The factual basis for this allegation is that at least one of the patents-in-suit was cited during the U.S. patent prosecution of patents assigned to Amazon itself, as well as patents assigned to Evi Technologies Limited and True Knowledge, companies whose technology was allegedly incorporated into Alexa after acquisition by Amazon (Compl. ¶9). This allegation suggests Plaintiff will argue Amazon had pre-suit knowledge of the patent portfolio.
VII. Analyst’s Conclusion: Key Questions for the Case
- A central issue will be one of architectural equivalence: the patents-in-suit, with a 1994 priority date, describe a specific and highly detailed system architecture for natural language processing. The key question for the court will be whether the complaint provides sufficient factual allegations to demonstrate that Amazon's modern, cloud-based Alexa system—built from acquired technologies like Evi and graph databases—implements the same particular data structures and processing steps recited in the claims, or if it represents a technologically distinct approach to achieving a similar outcome.
- The case will likely involve a critical question of definitional scope: can foundational claim terms rooted in the context of 1990s computer science, such as "dictionary data base" and "word sense numbers", be construed broadly enough to encompass modern knowledge graphs and entity-relationship models? The degree to which these terms are tethered to the specific embodiments described in the patents could be dispositive for infringement.
- A key procedural question will be one of evidentiary sufficiency: does the complaint, which relies on publicly available developer documents, technical articles, and marketing materials, allege sufficient facts to plausibly map the proprietary inner workings of the Alexa software to the detailed limitations of the asserted claims, thereby satisfying the pleading standards of Twombly and Iqbal?