Trainable Methods for Surface Natural Language Generation

By Adwait Ratnaparkhi. 2000

147 citations recorded by the source corpus.

We present three systems for surface natural language generation that are trainable from annotated corpora. The first two systems, called NLG1 and NLG2, require a corpus marked only with domainspecific semantic attributes, while the last system, called NLG3, requires a corpus marked with both semantic attributes and syntactic dependency information. All systems attempt to produce a grammatical natural language phrase from a domain-specific semantic representation. NLG1 serves a baseline system and uses phrase frequencies to generate a whole phrase in one step, while NLG2 and NLG3 use maximum entropy probability models to individually generate each word in the phrase. The systems NLG2 and NLG3 learn to determine both the word choice and the word order of the phrase. We present experiments in which we generate phrases to describe flights in the air travel domain.

Subfields: Artificial Intelligence, Natural Language Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Natural Language Processing

Citations in the corpus

Creating Training Corpora for NLG Micro-Planners (2017, methodology): “For the air travel domain, Ratnaparkhi (2000) created a dataset consisting of 5,426 data-text pairs with a richer vocabulary (927 words) and in the biology domain, the KBGen shared task (Banik et al., 2013) made available 284 data-text pairs where the data was extracted from an existing knowledge…”
From captions to visual concepts and back (2014, methodology): “During the generation stage, we perform a left-to-right beam search similar to the one used in [36].”
Survey of the State of the Art in Natural Language Generation: Core tasks, applications and evaluation (2017, methodology): “…not incorporate fluency heuristics, the decoder is interleaved with two further sources of linguistic knowledge by Konstas and Lapata (2013): (a) a weighted finite-state automaton (representing an n-gram language model); and (b) a dependency model (cf. Ratnaparkhi, 2000, , also discussed above).”
Strategies for Structuring Story Generation (2019, background): “sentence aggregation and lexicalization, and finally steps to resolve referring expressions (Hovy, 1990; Dalianis and Hovy, 1993; Wahlster et al., 1993; Ratnaparkhi, 2000; Malouf, 2000).”
Data-Driven Response Generation in Social Media (2011, background): “Additionally, there has been work on generating more natural utterances in goal-directed dialogue systems (Ratnaparkhi, 2000; Rambow et al., 2001).”
Multimodal Machine Learning: A Survey and Taxonomy (2017, background): “Language generation has been explored for a long time [177], with a lot of recent attention for tasks such as image and video description [20].”