Method and apparatus for lightweight multilingual natural language realizer

• US 10,445,432 B1
• Filed: 08/31/2016
• Issued: 10/15/2019
• Est. Priority Date: 08/31/2016
• Status: Active Grant

• Current Assignee: Arria Data2text Limited
• Original Assignee: Arria Data2text Limited
• Inventors: Reiter, Ehud Baruch, Turner, Ross John, Bradshaw, William Anthony, Viethen, Henriette Anna Elisabeth
• Primary Examiner(s): Hang, Vu B
• Application Number:
  US15/252,939
• Time in Patent Office: 1,140 Days
• CPC Class Codes: G06F 40/14 Tree-structured documents parsing G06F40/205; validation G06F40/226
• CPC Class Codes: G06F 40/154 Tree transformation for tree-structured or markup documents, e.g. XSLT, XSL-FO or stylesheets
• CPC Class Codes: G06F 40/211 Syntactic parsing, e.g. based on context-free grammar [CFG] or unification grammars
• CPC Class Codes: G06F 40/268 Morphological analysis
• CPC Class Codes: G06F 40/289 Phrasal analysis, e.g. finite state techniques or chunking
• CPC Class Codes: G06F 40/30 Semantic analysis
• CPC Class Codes: G06F 40/56 Natural language generation

Abstract

Methods, apparatuses, and computer program products are described herein that are configured to be embodied as a natural language realizer in the form of a lightweight multilingual realizer. In some example embodiments, a method is provided that comprises generating syntactically correct natural language. The method of this embodiment may also include at a syntax processor, propagating features among a group of words, and adding closed class function words and morphological inflections.

Claims

1. A method for generating syntactically correct natural language, the method comprising:

accessing, using processing circuitry, a realization element data structure, wherein the realization element data structure is a syntactic representation of phrases and sentences; and

accessing, using the processing circuitry, at least one syntax rule file at a syntax processor in a natural language realizer, wherein the at least one syntax rule file is configured for a language;

generating, using the processing circuitry, a syntactically processed realization element data structure, at the syntax processor, using the at least one syntax rule file in the syntax processor by;

propagating, using the processing circuitry, features in the realization element data structure by comparing one or more realization elements in the realization element data structure;

adding, using the processing circuitry and to the realization element data structure, closed class function words and morphological inflections by utilizing rules from the syntax rule file to determine the closed class function words and morphological inflections to add to base words and phrases in the realization element data structure; and

outputting, using the processing circuitry and for display via a computing device, natural language text generated based on the realization element data structure.

2. The method of claim 1, wherein the syntax processor comprises one or more language specific syntax processors, which accesses and applies rules from the syntax rule file.

3. The method of claim 1, wherein adding closed class function words and morphological inflections comprises:

accessing, using the processing circuitry, rewrite production rules for adding closed class function words and morphological inflections in the at least one syntax rule file;

determining, using the processing circuitry, base words and phrases of the elements of the realization element data structure to be modified by the addition of closed class function words and morphological inflections, wherein the realization elements include determined features;

applying, using the processing circuitry, the rewrite production rules for adding close class function words and morphological inflections by iteratively processing the realization elements according to the determined features and corresponding closed class function word rule, by modifying the words and phrases and adding the determined closed class function words and morphological inflections; and

determining, using the processing circuitry, if all applicable rules for adding closed class function words and morphological inflections have been applied and returning the realization element data structure for further use by the natural language realizer.

4. The method of claim 1, wherein propagating features comprises:

applying propagation and unification rules by;

determining, using the processing circuitry, which words from the elements of the data structure to unify by comparing the elements of the data structure with rules for feature unification, and

unifying, using the processing circuitry, features among the elements of the data structure by modifying the determined words according to the rules for feature unification.

5. The method of claim 1, wherein the realization element data structure is received from either a microplanner or a lightweight NLG service invoked from a template.

6. The method of claim 1, wherein the natural language realizer further optionally comprises a morphology processor, a morphophonology processor, an orthography processor, and a mark-up processor, and wherein the natural language realizer outputs fully realized natural language text.

7. The method of claim 6, wherein the morphology processor, the morphophonology processor, the orthography processor, and the mark-up processor each comprise a language specific function processor, which accesses and applies rules from a language specific rule file.

8. The method of claim 6, further comprising generating, using the processing circuitry, a processed realization element data structure, at the morphophonology processor, using the at least one language morphophonology rule file in the language morphophonology processor to generate a morphophonologically processed realization element data structure by:

determining, using the processing circuitry, morphophonological features to modify the elements of the data structure according to the language morphophonology rule file; and

applying, using the processing circuitry, the determined features to the realization elements.

9. The method of claim 6, further comprising generating, using the processing circuitry, a processed realization element data structure, at the orthography processor, using the at least one language orthography style file in the language orthography processor to generate an orthographically processed realization element data structure by at least one of:

capitalizing individual words;

formatting numbers such as applying currency or date formats;

performing punctuation operations;

or

placing white space characters between words.

10. The method of claim 6, further comprising generating, using the processing circuitry, a processed realization element data structure, at the mark-up processor, using the at least one language mark-up rule file in the language mark-up processor to generate a marked-up realization element data structure by at least one of:

converting internal markups found in the realization element data structure into a mark-up format or into a plurality of formats that can be understood by a plurality of natural language text presentation systems.

11. The method of claim 1, wherein the natural language realizer receives language specific rule sets, wherein the language specific rule sets comprise:

a syntax rule file, including at least rules to add closed class function words and morphological inflections, and propagate features; and

optionally comprises any of;

a language morphology rule file;

a language morphophonology rule file;

a language orthography style file; and

a language mark-up rule file.

12. An apparatus comprising:

at least one processor; and

at least one memory including computer program code, the at least one memory and the computer program code configured to, with the at least one processor, cause the apparatus to at least;

access a realization element data structure, wherein the realization element data structure is a syntactic representation of phrases and sentences; and

access at least one syntax rule file at a syntax processor in a natural language realizer, wherein the at least one syntax rule file is configured for a language;

generate a syntactically processed realization element data structure, at the syntax processor, using the at least one syntax rule file in the syntax processor by;

propagating features in the realization element data structure by comparing one or more realization elements in the realization element data structure;

adding closed class function words and morphological inflections to the realization element data structure by utilizing rules from the syntax rule file to determine closed class function words and morphological inflections to add to base words and phrases in the realization element data structure; and

output, for display via a computing device, natural language text generated based on the realization element data structure.

13. A computer program product comprising:

at least one computer readable non-transitory memory medium having program code instructions stored thereon, the program code instructions which when executed by an apparatus cause the apparatus at least to;

access a realization element data structure, wherein the realization element data structure is a syntactic representation of phrases and sentences; and

access at least one syntax rule file at a syntax processor in a natural language realizer, wherein the at least one syntax rule file is configured for a language;

generate a syntactically processed realization element data structure, at the syntax processor, using the at least one syntax rule file in the syntax processor by;

propagating features in the realization element data structure by comparing one or more realization elements in the realization element data structure;

adding closed class function words and morphological inflections to the realization element data structure by utilizing rules from the syntax rule file to determine the closed class function words and morphological inflections to add to base words and phrases in the realization element data structure; and

output, for display via a computing device, natural language text generated based on the realization element data structure.