Conversational syntax using constrained natural language processing to access datasets

Abstract

Generally, techniques are described for various aspects of accessing a dataset. An apparatus includes a memory configured to store the dataset and a processor configurable to execute the techniques. The processor can expose language subsurfaces that specify a natural language containment hierarchy that defines a grammar for a natural language as a hierarchical arrangement of multiple language subsurfaces. The processor can receive a query for accessing the dataset that conforms to portions of the natural language provided by the exposed language subsurfaces. The processor can convert the query into one or more statements that conform to a formal syntax associated with the dataset, access the dataset to obtain query results based on the one or more statements, and output the query results.

Claims

1. A device configured to access a dataset, A memory configured to store the aforementioned dataset, It comprises one or more processors, and the one or more processors are By identifying a natural language inclusion hierarchy that defines grammar for natural language as a hierarchical arrangement of multiple language subsurfaces, and exposing language subsurfaces, To access the dataset, receive queries that conform to the portion of natural language provided by the exposed language subsurface, Convert the aforementioned query into one or more statements that conform to the formal syntax associated with the aforementioned dataset, Based on the one or more statements mentioned above, access the dataset to obtain query results, The system is configured to output the aforementioned query results, The one or more processors described above are: To access the dataset, receive an additional query that identifies dimensions in the dataset that are not present in the query results, in accordance with the portion of the natural language provided by the exposed language subsurface. The identified dimension is determined not to exist in the query result. Convert the aforementioned additional query into one or more additional statements conforming to the aforementioned formal syntax, Based on the one or more additional statements and in response to determining that the identified dimension does not exist in the query results, access the dataset to retrieve the additional query results. The apparatus is further configured to output the additional query results along with a statement indicating that the additional query results were obtained from the dataset and not from the query results.

2. A device configured to access a dataset, A memory configured to store the aforementioned dataset, It comprises one or more processors, and the one or more processors are By identifying a natural language inclusion hierarchy that defines grammar for natural language as a hierarchical arrangement of multiple language subsurfaces, and exposing language subsurfaces, To access the dataset, receive queries that conform to the portion of natural language provided by the exposed language subsurface, Convert the aforementioned query into one or more statements that conform to the formal syntax associated with the aforementioned dataset, Based on the one or more statements mentioned above, access the dataset to obtain query results, The system is configured to output the aforementioned query results, The aforementioned query includes a multipart query having multiple query statements, The exposed language subsurface removes ambiguity when defining the query, so that the multiple query statements forming the multipart query can be defined in any order. The apparatus is characterized in that the one or more processors are configured to convert the multiple partial queries into the same one or more statements, regardless of the order in which the multiple query statements are defined, in order to form the multiple partial queries.

3. A device configured to access a dataset, A memory configured to store the aforementioned dataset, It comprises one or more processors, and the one or more processors are By identifying a natural language inclusion hierarchy that defines grammar for natural language as a hierarchical arrangement of multiple language subsurfaces, and exposing language subsurfaces, To access the dataset, receive queries that conform to the portion of natural language provided by the exposed language subsurface, Convert the aforementioned query into one or more statements that conform to the formal syntax associated with the aforementioned dataset, Based on the one or more statements mentioned above, access the dataset to obtain query results, The system is configured to output the aforementioned query results, The aforementioned dataset is one of several datasets, The one or more processors described above are: Identify the relationships between one or more dimensions of the aforementioned multiple datasets, Based on the identified relationships, a graph data structure is generated having nodes representing each of the multiple datasets and edges representing the relationships between one or more dimensions of the multiple datasets. Based on the query, the graph data structure is traversed to identify the shortest path through the graph data structure to satisfy the query. To obtain a combined dataset, the dataset is automatically combined with one or more additional datasets of the multiple datasets identified along the shortest path. The apparatus is further configured to access the combined dataset in order to obtain the query results based on one or more of the statements.

4. The one or more processors described above are: Based on the aforementioned query, in order to identify additional paths through the graph data structure that satisfy the aforementioned query, the graph data structure is traversed, The apparatus according to claim 3, further configured to output a display identifying the additional route through the graph data structure.

5. The apparatus according to claim 4, characterized in that the display is a link to an improved query which results in traversing the additional paths through the graph data structure.

6. The apparatus according to any one of claims 1 to 5, characterized in that the formal syntax includes a structured query language syntax or a Pandas data frame syntax.

7. A method for accessing a dataset, One or more processors expose language subsurfaces that identify a natural language inclusion hierarchy that defines a grammar for a natural language as a hierarchical arrangement of multiple language subsurfaces, The one or more processors receive queries that conform to the portion of natural language provided by the exposed language subsurface for accessing the dataset, The one or more processors translate the query into one or more statements that conform to the formal syntax associated with the dataset, The one or more processors access the dataset to obtain query results based on the one or more statements, The one or more processors output the query results, The one or more processors receive an additional query for accessing the dataset, which conforms to the portion of the natural language provided by the exposed language subsurface and identifies dimensions in the dataset that are not present in the query results, The one or more processors determine that the identified dimension does not exist in the query result, The one or more processors convert the additional query into one or more additional statements conforming to the formal syntax, The one or more processors access the dataset to retrieve additional query results based on the one or more additional statements and in response to determining that the identified dimension does not exist in the query results, The method further comprises the one or more processors outputting the additional query results along with a statement indicating that the additional query results were obtained from the dataset and not from the query results.

8. A method for accessing a dataset, One or more processors expose language subsurfaces that identify a natural language inclusion hierarchy that defines a grammar for a natural language as a hierarchical arrangement of multiple language subsurfaces, The one or more processors receive queries that conform to the portion of natural language provided by the exposed language subsurface for accessing the dataset, The one or more processors translate the query into one or more statements that conform to the formal syntax associated with the dataset, The one or more processors access the dataset to obtain query results based on the one or more statements, The one or more processors output the query results, The aforementioned query includes a multipart query having multiple query statements, The exposed language subsurface removes ambiguity when defining the query, so that the multiple query statements forming the multipart query can be defined in any order. The method for converting the query is characterized in that one or more processors convert the multipart query into the same one or more statements, regardless of the order in which the multiple query statements are defined, in order to form the multipart query.

9. A method for accessing a dataset, One or more processors expose language subsurfaces that identify a natural language inclusion hierarchy that defines a grammar for a natural language as a hierarchical arrangement of multiple language subsurfaces, The one or more processors receive queries that conform to the portion of natural language provided by the exposed language subsurface for accessing the dataset, The one or more processors translate the query into one or more statements that conform to the formal syntax associated with the dataset, The one or more processors access the dataset to obtain query results based on the one or more statements, The one or more processors output the query results, The aforementioned dataset is one of several datasets, The aforementioned method, The one or more processors identify the relationships between one or more dimensions of the plurality of datasets, The one or more processors generate a graph data structure having nodes representing each of the multiple datasets and edges representing the relationships between one or more dimensions of the multiple datasets, based on the identified relationships. The one or more processors traverse the graph data structure to identify the shortest path through the graph data structure to satisfy the query, The one or more processors automatically combine the dataset with one or more additional datasets of the plurality of datasets identified along the shortest path in order to obtain a combined dataset. A method characterized in that the one or more processors further access the combined dataset to obtain the query results based on the one or more statements.

10. The one or more processors traverse the graph data structure to identify additional paths through the graph data structure that satisfy the query, The method according to 9, further comprising the one or more processors outputting a display that identifies the additional path through the graph data structure.

11. The method according to 10, characterized in that the display is a link to an improved query which results in traversing the additional paths through the graph data structure.

12. The method according to any one of claims 7 to 11, characterized in that the formal syntax includes a structured query language syntax or a Pandas dataframe syntax.

13. A non-temporary computer-readable storage medium storing instructions, wherein when an instruction is executed, it is sent to one or more processors. By identifying a natural language inclusion hierarchy that defines grammar for natural language as a hierarchical arrangement of multiple language subsurfaces, and exposing language subsurfaces, To access the dataset, it receives queries that conform to the portion of natural language provided by the exposed language subsurface, Convert the aforementioned query into one or more statements that conform to the formal syntax associated with the aforementioned dataset. Based on one or more of the above statements, access the dataset to obtain query results, Furthermore, by one or more of the aforementioned processors, To access the dataset, receive an additional query that identifies dimensions in the dataset that are not present in the query results, in accordance with the portion of the natural language provided by the exposed language subsurface. The identified dimension is determined not to exist in the query result. Convert the aforementioned additional query into one or more additional statements conforming to the aforementioned formal syntax, Based on the one or more additional statements, and in response to determining that the identified dimension does not exist in the query results, access the dataset to retrieve the additional query results. A non-temporary computer-readable storage medium characterized in that it outputs the additional query results along with a statement indicating that the additional query results were obtained from the dataset and not from the query results.

14. A non-temporary computer-readable storage medium storing instructions, wherein when an instruction is executed, it is sent to one or more processors. By identifying a natural language inclusion hierarchy that defines grammar for natural language as a hierarchical arrangement of multiple language subsurfaces, and exposing language subsurfaces, To access the dataset, it receives queries that conform to the portion of natural language provided by the exposed language subsurface, Convert the aforementioned query into one or more statements that conform to the formal syntax associated with the aforementioned dataset. Based on one or more of the above statements, access the dataset to obtain query results, The aforementioned query includes a multipart query having multiple query statements, The exposed language subsurface removes ambiguity when defining the query, so that the multiple query statements forming the multipart query can be defined in any order. A non-temporary computer-readable storage medium characterized in that, in order to form the multipart query, one or more processors convert the multipart query into the same one or more statements, regardless of the order in which the multiple query statements are defined.

15. A non-temporary computer-readable storage medium storing instructions, wherein when an instruction is executed, it is sent to one or more processors. By identifying a natural language inclusion hierarchy that defines grammar for natural language as a hierarchical arrangement of multiple language subsurfaces, and exposing language subsurfaces, To access the dataset, it receives queries that conform to the portion of natural language provided by the exposed language subsurface, Convert the aforementioned query into one or more statements that conform to the formal syntax associated with the aforementioned dataset. Based on one or more of the above statements, access the dataset to obtain query results, The aforementioned dataset is one of several datasets, Furthermore, by one or more of the aforementioned processors, To identify the relationships between one or more dimensions of the aforementioned multiple datasets, Based on the identified relationships, a graph data structure is generated having nodes representing each of the multiple datasets and edges representing the relationships between one or more dimensions of the multiple datasets. Based on the query, the graph data structure is traversed to identify the shortest path through the graph data structure to satisfy the query. To obtain a combined dataset, the dataset is automatically combined with one or more additional datasets of the multiple datasets identified along the shortest path. A non-temporary computer-readable storage medium characterized by allowing access to the combined dataset to obtain the query results based on one or more of the aforementioned statements.

Images