In many cases, splitting occurs when an entity takes part in two (or more) distinct events described in a single sentence. For example, in Sentence (1), bricks is involved in two events: “being resistant to cold” and “enabling the construction of permanent buildings.”

(1) Original: Being more resistant to cold, bricks enabled the construction of
permanent buildings.

(2) Simplified: Bricks were more resistant to cold. Bricks enabled the
construction of permanent buildings.

Even though deciding when and where to split can be determined by syntax, constructing the second sentence in the split by adding the shared element with the first split should be accomplished by semantic information, because we need to identify the entity involved in both events.

University of Michigan - Ann Arbor

- Monolingual Statistical Machine Translation

- Grammar Induction

- Semantics-Assisted

Data-Driven Approaches to Sentence Simplification

Alva-Manchego, F., Scarton, C., & Specia, L. (2020). Data-driven sentence simplification: Survey and benchmark. Computational Linguistics, 46(1), 135-187.

Data-Driven Sentence Simplification: Survey and Benchmark

Given a sentence f in the source language, the goal of an SMT model is to produce a translation e in the target language. This is modeled using the noisy channel framework Equation (9).

This framework relies on a translation model p(f|e) and a language model p(e).
In addition, a decoder is in charge of producing the most probable e given an f. 

There are different approaches for implementing the translation model and the decoder. In practice, they all rely on a linear combination of these and additional features, which are directly optimized to maximize translation quality, rather than on the generative noisy channel model.

Monolingual Statistical Machine Translation

In this approach, SS is modeled as a tree-to-tree rewriting problem. Approaches typically follow a two-step process: 

(1) use parallel corpora of aligned original simplified sentence pairs to extract a set of tree transformation rules, and then

(2) learn how to select which rule(s) to apply to unseen sentences to generate the best simplified output. 

This is analogous to how a SBSMT approach works: The rules would be the features, and the decoder applies the learned model deciding how to use these rules.

Grammar Induction

Semantics-Assisted

In this approach, SS is modeled as a sequence-to-sequence problem, and tackled
normally with an attention-based encoder-decoder architecture. The encoder projects the source sentence into a set of continuous vector representations from which the decoder generates the target sentence. 

A major advantage of this approach is that it allows training of end-to-end models without needing to extract features or estimate individual model components, such as the language model. In addition, all simplification transformations can be learned simultaneously, instead of developing individual mechanisms as in previous research.

Learn Before

Related