Technologies for valid dependency parsing in casual text
Technologies for natural language processing include a computing device that loads natural language text data that includes multiple words, such as a naturally elicited response or comment data. The computing device applies a constituency parser to the natural language text data to generate a constituency parse tree. The constituency parse tree may include multiple nodes that each correspond to a hierarchical constituent of the natural language text. The computing device applies a constituency rule to identify clauses in the natural language text data. The constituency rule may identify a left-most daughter of each subordinate clause node of the constituency parse tree. Each subordinate clause is distinct, preventing dependencies from crossing clause boundaries. The computing device applies a dependency parser to each clause to generate a corresponding dependency parse. Each dependency parse may include a graph with nodes corresponding to words of the clause and edges corresponding to dependency relationships.
1. A computing device for natural language parsing, the computing device comprising:
input logic to load natural language text data, wherein the natural language text data comprises a plurality of words;
constituency parser logic to apply a constituency parser to the natural language text data to generate a constituency parse tree;
clause boundary logic to apply a constituency rule to the constituency parse tree to identify a plurality of clauses in the natural language text data by determining one or more clause bounds, wherein each of the plurality of clauses comprises a disjoint subset of the plurality of words of the natural language text data; and
dependency parser logic to apply a dependency parser to each of the plurality of clauses to generate a dependency parse of each of the plurality of clauses.
2. The computing device of claim 1 , wherein the constituency parse tree comprises a plurality of nodes, wherein each of the plurality of nodes corresponds to a hierarchical constituent of the natural language text data.
3. The computing device of claim 2 , wherein to apply the constituency rule comprises to identify a left-most daughter of a subordinate clause node of the constituency parse tree.
4. The computing device of claim 1 , wherein each dependency parse comprises a graph that includes a plurality of nodes and one or more edges, wherein each of the plurality of nodes corresponds to a word of a corresponding clause, and wherein each of the one or more edges corresponds to a dependency relationship within the corresponding clause.
5. The computing device of claim 1 , wherein the natural language text data comprises a naturally elicited response.
6. The computing device of claim 1 , wherein the natural language text data comprises comment data submitted by a user to a website.
7. The computing device of claim 1 , wherein to apply the constituency parser comprises to apply a constituency parser that does not rely on noun chunking as an initial parsing step.
8. The computing device of claim 1 , wherein to apply the constituency rule comprises to determine a plurality of clause boundaries, wherein each of the plurality of clause boundaries is associated with a corresponding clause of the plurality of clauses.
9. The computing device of claim 8 , wherein the plurality of clause boundaries comprises a plurality of non-overlapping character offsets in the natural language text data, wherein each of the plurality of non-overlapping character offsets identifies a corresponding word of the natural language text data.
10. One or more non-transitory, computer readable media comprising a plurality of instructions stored thereon that in response to being executed cause a computing device to:
load natural language text data, wherein the natural language text data comprises a plurality of words;
apply a constituency parser to the natural language text data to generate a constituency parse tree;
apply a constituency rule to the constituency parse tree to identify a plurality of clauses in the natural language text data by determining one or more clause bounds, wherein each of the plurality of clauses comprises a disjoint subset of the plurality of words of the natural language text data; and
apply a dependency parser to each of the plurality of clauses to generate a dependency parse of each of the plurality of clauses.
11. The one or more non-transitory, computer readable media of claim 10 , wherein the constituency parse tree comprises a plurality of nodes, wherein each of the plurality of nodes corresponds to a hierarchical constituent of the natural language text data.
12. The one or more non-transitory, computer readable media of claim 11 , wherein to apply the constituency rule comprises to identify a left-most daughter of a subordinate clause node of the constituency parse tree.
13. The one or more non-transitory, computer readable media of claim 10 , wherein each dependency parse comprises a graph that includes a plurality of nodes and one or more edges, wherein each of the plurality of nodes corresponds to a word of a corresponding clause and wherein each of the one or more edges corresponds to a dependency relationship within the corresponding clause.
14. The one or more non-transitory, computer readable media of claim 10 , wherein to apply the constituency rule comprises to determine a plurality of clause boundaries, wherein each of the plurality of clause boundaries is associated with a corresponding clause of the plurality of clauses.
15. The one or more non-transitory, computer readable media of claim 14 , wherein the plurality of clause boundaries comprises a plurality of non-overlapping character offsets in the natural language text data, wherein each of the plurality of non-overlapping character offsets identifies a corresponding word of the natural language text data.
16. A method for natural language parsing, the method comprising:
loading, by a computing device, natural language text data, wherein the natural language text data comprises a plurality of words;
applying, by the computing device, a constituency parser to the natural language text data to generate a constituency parse tree;
applying, by the computing device, a constituency rule to the constituency parse tree to identify a plurality of clauses in the natural language text data by determining one or more clause bounds, wherein each of the plurality of clauses comprises a disjoint subset of the plurality of words of the natural language text data; and
applying, by the computing device, a dependency parser to each of the plurality of clauses to generate a dependency parse of each of the plurality of clauses.
17. The method of claim 16 , wherein the constituency parse tree comprises a plurality of nodes, wherein each of the plurality of nodes corresponds to a hierarchical constituent of the natural language text data.
18. The method of claim 17 , wherein applying the constituency rule comprises identifying a left-most daughter of a subordinate clause node of the constituency parse tree.
19. The method of claim 16 , wherein each dependency parse comprises a graph that includes a plurality of nodes and one or more edges, wherein each of the plurality of nodes corresponds to a word of a corresponding clause, and wherein each of the one or more edges corresponds to a dependency relationship within the corresponding clause.
20. The method of claim 16 , wherein applying the constituency rule comprises determining a plurality of clause boundaries, wherein each of the plurality of clause boundaries is associated with a corresponding clause of the plurality of clauses.