SLING: A framework for frame semantic parsing

By Michael Ringgaard, Rahul Gupta, Fernando C Pereira. 2017

55 citations recorded by the source corpus.

We describe SLING, a framework for parsing natural language into semantic frames. SLING supports general transition-based, neural-network parsing with bidirectional LSTM input encoding and a Transition Based Recurrent Unit (TBRU) for output decoding. The parsing model is trained end-to-end using only the text tokens as input. The transition system has been designed to output frame graphs directly without any intervening symbolic representation. The SLING framework includes an efficient and scalable frame store implementation as well as a neural network JIT compiler for fast inference during parsing. SLING is implemented in C++ and it is available for download on GitHub.

Subfields: Natural Language Processing, Computational Linguistics & Semantics

How researchers see this paper

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

In Natural Language Processing

In Computational Linguistics & Semantics

Citations in the corpus

Entities as Experts: Sparse Memory Access with Entity Supervision (2020, methodology): “When the answer is an entity, in our entity vocabulary, we link the answer string to an entity ID using the SLING phrase table 8 (Ringgaard et al., 2017).”
Differentiable Reasoning over a Virtual Knowledge Base (2020, methodology): “For the multi-hop slot-filling experiments below, we used WikiData (Vrandečić & Krötzsch, 2014) as our KB, Wikipedia as the corpus, and SLING (Ringgaard et al., 2017) to identify entity mentions.”
X-FACTR: Multilingual Factual Knowledge Retrieval from Pretrained Language Models (2020, methodology): “Given a pair of languages, we ﬁrst identify Wikipedia sentences that mention entities from our benchmark using SLING (Ringgaard et al., 2017).”
Paraphrase Augmented Task-Oriented Dialog Generation (2020, methodology): “In this structure, a semantic parser SLING (Ringgaard et al., 2017) is used to analyze the semantic frame of an utterance and the semantic role of each token in it.”
Robustness in Coreference Resolution (2020, background): “Ringgaard et al. (2017). We show that traditional features, which are obtained from error-prone pipelined modules, encode auxiliary information to what is captured by the learned representations, and their incorporation results in significant improvements.”
VQA With No Questions-Answers Training (2018, background): “Understanding natural language questions and parsing them to a logical form is a hard problem, still under study [29, 7, 74, 11, 58].”
A Unified Encoder-Decoder Framework with Entity Memory (2022, methodology): “If an answer can be directly linked to a Wikipedia entity according to Google’s SLING (Ringgaard et al., 2017) phrase table, it is counted as an entity answer, otherwise a non-entity answer.”