Architecture and processes for computer learning and understanding
First Claim
1. A computer-implemented method, comprising:
- maintaining a collection of frame structures within memory of a computing system, individual frame structures of the collection of frame structures defining a set of roles for one or more generative semantic primitive (GSP) structures that are commonly associated with a context of a setting in natural language stories;
receiving, by the computing system, a natural language story;
performing, by the computing system, a syntactic analysis of the natural language story to produce one or more linguistic analysis results, wherein producing the one or more linguistic analysis results includes;
performing analysis on the natural language story to produce a syntactic representation of the natural language story;
performing predicate argument structure (PAS) analysis on the syntactic representation of the natural language story;
assigning one or more entity types to one or more words in the natural language story; and
determining co-reference chains in the one or more words in the natural language story;
creating, by the computing system, a new frame structure that defines a new set of roles for one or more new GSP structures that account for the one or more linguistic analysis results derived from the natural language story, a GSP structure of the one or more new GSP structures having at least a threshold probability of being included in the new frame structure that is associated with a particular context of a particular setting in the natural language story, wherein creating the new frame structure comprises replacing specific entities in the natural language story as found in the one or more linguistic analysis results with corresponding generic entity types and aligning, by the computing system, the specific entities in the natural language story to the new set of roles defined in the new frame structure; and
storing the new frame structure as part of the collection of frame structures in the memory of the computing system.
2 Assignments
0 Petitions
Accused Products
Abstract
An architecture and processes enable computer learning and developing an understanding of arbitrary natural language text through collaboration with humans in the context of joint problem solving. The architecture ingests the text and then syntactically and semantically processes the text to infer an initial understanding of the text. The initial understanding is captured in a story model of semantic and frame structures. The story model is then tested through computer generated questions that are posed to humans through interactive dialog sessions. The knowledge gleaned from the humans is used to update the story model as well as the computing system'"'"'s current world model of understanding. The process is repeated for multiple stories over time, enabling the computing system to grow in knowledge and thereby understand stories of increasingly higher reading comprehension levels.
-
Citations
20 Claims
-
1. A computer-implemented method, comprising:
-
maintaining a collection of frame structures within memory of a computing system, individual frame structures of the collection of frame structures defining a set of roles for one or more generative semantic primitive (GSP) structures that are commonly associated with a context of a setting in natural language stories; receiving, by the computing system, a natural language story; performing, by the computing system, a syntactic analysis of the natural language story to produce one or more linguistic analysis results, wherein producing the one or more linguistic analysis results includes; performing analysis on the natural language story to produce a syntactic representation of the natural language story; performing predicate argument structure (PAS) analysis on the syntactic representation of the natural language story; assigning one or more entity types to one or more words in the natural language story; and determining co-reference chains in the one or more words in the natural language story; creating, by the computing system, a new frame structure that defines a new set of roles for one or more new GSP structures that account for the one or more linguistic analysis results derived from the natural language story, a GSP structure of the one or more new GSP structures having at least a threshold probability of being included in the new frame structure that is associated with a particular context of a particular setting in the natural language story, wherein creating the new frame structure comprises replacing specific entities in the natural language story as found in the one or more linguistic analysis results with corresponding generic entity types and aligning, by the computing system, the specific entities in the natural language story to the new set of roles defined in the new frame structure; and storing the new frame structure as part of the collection of frame structures in the memory of the computing system. - View Dependent Claims (2, 3, 4, 5)
-
-
6. A computer-implemented method, comprising:
-
maintaining a collection of frame structures within memory of a computing system, individual frame structures of the collection of frame structures defining a set of roles and relationships among the set of roles for one or more generative semantic primitive (GSP) structures that are commonly associated with a particular context of a setting or a background that may be expressed in natural language stories, a GSP structure of the one or more GSP structures having at least a threshold probability of being included in the individual frame structures that is associated with the particular context of the setting or the background of the natural language stories; processing, by the computing system, language corpora to produce knowledge resources indicative of natural language usage; creating, by the computing system, a new frame structure that defines a new set of roles and new relationships among the new set of roles based at least in part on the knowledge resources; engaging in multiple dialog sessions with computing devices of multiple human users in parallel to receive responses from the multiple human users for use by the computing system to evaluate, by the computing system, an extent to which the new frame structure predicts natural language excerpts in the language corpora based at least in part on mapping one or more entities in the natural language excerpts to one or more roles of the new set of roles defined in the new frame structure; and storing the new frame structure as part of the collection of frame structures in the memory of the computing system. - View Dependent Claims (7, 8, 9, 10, 11)
-
-
12. A computer-implemented method, comprising:
-
maintaining a collection of frame structures within memory of a computing system, individual frame structures of the collection of frame structures defining a set of roles and relationships among the set of roles for multiple generative semantic primitive (GSP) structures that are commonly associated with a particular context of a setting or a background in natural language stories, individual GSP structures of the multiple GSP structures having at least a threshold probability of being included in the individual frame structures that is associated with the particular context of the setting or the background in the natural language stories; processing, by the computing system, the collection of frame structures stored in the memory according to one of deductive or abductive processes; creating, by the computing system, a new frame structure through logical and probabilistic reasoning of results from processing the collection of frame structures stored in the memory; engaging in multiple dialog sessions with computing devices of multiple human users in parallel to receive responses from the multiple human users for use by the computing system to evaluate, by the computing system, an extent to which the new frame structure predicts natural language excerpts in language corpora based at least in part on mapping one or more entities in the natural language excerpts to one or more roles defined in the new frame structure; and storing the new frame structure as part of the collection of frame structures in the memory of the computing system. - View Dependent Claims (13, 14, 15)
-
-
16. A computing system, comprising:
-
a datastore to store a collection of frame structures, individual frame structures of the collection of frame structures defining a set of roles for one or more generative semantic primitive (GSP) structures that are commonly associated with a particular context of a setting or a background in natural language stories; one or more processors; and memory coupled to the one or more processors, the memory storing; a story parsing engine to syntactically analyze a natural language story to produce linguistic analysis results, wherein producing the linguistic analysis results includes; performing analysis on the natural language story to produce a syntactic representation of the natural language story; performing predicate argument structure (PAS) analysis on the syntactic representation of the natural language story; assigning one or more entity types to one or more words in the natural language story; and determining co-reference chains in the one or more words in the natural language story; a knowledge induction engine to process language corpora to produce knowledge resources indicative of natural language usage; and a knowledge integration engine to form a new frame structure based at least in part on at least one of the linguistic analysis results or the knowledge resources and based at least in part on expanding at least one frame structure of the collection of frame structures to include a new GSP structure, wherein the knowledge integration engine is configured to replace specific entities in the natural language story as found in the linguistic analysis results with corresponding generic entity types and to align specific entities in the natural language story to a new set of roles defined in the new frame structure, wherein the new frame structure is stored in the datastore. - View Dependent Claims (17, 18, 19, 20)
-
Specification