SEAGLE: A Platform for Comparative Evaluation of Semantic Encoders for Information Retrieval
2
Citation
20
Reference
10
Related Paper
Citation Trend
Abstract:
Fabian David Schmidt, Markus Dietsche, Simone Paolo Ponzetto, Goran Glavaš. Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP): System Demonstrations. 2019.A building is constructed by parts and joints. The joints, each other, have the definite relation. In other words, there is a system in these joints. We call it "Joint System". This is consisted of two systems, "Joint Location System" and "Joint Inner System". The former is where the joints in the building are, and the latter is what the relations between parts in a joint are. We studied "Joint Location System" only, by three aims; 1) To make the abstracted symbols which indicates distinctly the principles of "Joint Location System". 2) To make clear the principles of "Joint Location System" using the abstracted symbols. 3) To make the building construction system based on the principle of "Joint Location System". Contents 1. Aim of this study 2. The concept of "Joint Location System" 3. To make the abstracted symbols 4. Analytic method of "Joint Location System" 5. To analyze the buildings 6. To make the building construction system based on the principle of "Joint Location System" 7. Result
Cite
Citations (0)
Many military communication domains involve rapidly conveying situation awareness with few words. Converting natural language utterances to logical forms in these domains is challenging, as these utterances are brief and contain multiple intents. In this paper, we present a first effort toward building a weakly-supervised semantic parser to transform brief, multi-intent natural utterances into logical forms. Our findings suggest a new “projection and reduction” method that iteratively performs projection from natural to canonical utterances followed by reduction of natural utterances is the most effective. We conduct extensive experiments on two military and a general-domain dataset and provide a new baseline for future research toward accurate parsing of multi-intent utterances.
Logical form
Natural language understanding
Baseline (sea)
Cite
Citations (0)
To determine why the U.S. Department of Homeland Security (DHS) has struggled with translating jointly defined requirements into joint acquisition programs, one author referenced both current DHS policy and the experiences of recent joint programs. In addition to presenting his findings, he makes recommendations for guidance for future efforts.
Homeland
Cite
Citations (0)
Some writers consider a few joint development agreements should be copied as models by other states in subsequent joint development arrangements. Other writers believe that no joint development agreement should be a model for states owing to differences in circumstances. This article examines the agreements regarded as model joint development agreements, the body of literature on the issue, as well as one of the latest joint development agreements, in order to determine whether a model agreement is desirable for joint development and, if so, whether any of these agreements can be a model for other joint development agreements.
Cite
Citations (5)
Large language models can perform semantic parsing with little training data, when prompted with in-context examples. It has been shown that this can be improved by formulating the problem as paraphrasing into canonical utterances, which casts the underlying meaning representation into a controlled natural language-like representation. Intuitively, such models can more easily output canonical utterances as they are closer to the natural language used for pre-training. Recently, models also pre-trained on code, like OpenAI Codex, have risen in prominence. For semantic parsing tasks where we map natural language into code, such models may prove more adept at it. In this paper, we test this hypothesis and find that Codex performs better on such tasks than equivalent GPT-3 models. We evaluate on Overnight and SMCalFlow and find that unlike GPT-3, Codex performs similarly when targeting meaning representations directly, perhaps because meaning representations are structured similar to code in these datasets.
Natural language understanding
Code (set theory)
Representation
Cite
Citations (0)
Large language models can perform semantic parsing with little training data, when prompted with in-context examples. It has been shown that this can be improved by formulating the problem as paraphrasing into canonical utterances, which casts the underlying meaning representation into a controlled natural language-like representation. Intuitively, such models can more easily output canonical utterances as they are closer to the natural language used for pre-training. Recently, models also pre-trained on code, like OpenAI Codex, have risen in prominence. For semantic parsing tasks where we map natural language into code, such models may prove more adept at it. In this paper, we test this hypothesis and find that Codex performs better on such tasks than equivalent GPT-3 models. We evaluate on Overnight and SMCalFlow and find that unlike GPT-3, Codex performs similarly when targeting meaning representations directly, perhaps because meaning representations are structured similar to code in these datasets.
Natural language understanding
Code (set theory)
Representation
Semantic role labeling
Cite
Citations (37)
Bridge (graph theory)
Lap joint
Cite
Citations (0)
Texture (cosmology)
Cite
Citations (4)
Dividing the system of joint early-warning into inhomogeneous subsystems,the paper analyses the detecting efficiency of joint armaments by the means of system integration,linking the development requirement of air-defence forces joint early-warning system.It also provides the advice of establishing the joint early-warning system quantificationally.
Early warning system
Cite
Citations (0)
This paper describes a process of cutting blocks from statistically generated finite joint polygons in 3D space.If the ratio of joint length divided by joint spacing is less than 10,the rock mass is likely connected.If this joint length ratio is greater than 10,the rock is likely to be blocky.An algorithm is also presented for finding all removable blocks along any given moving direction.The rock mass boundary can be any excavated and natural free surfaces.The algorithm works for both joint sets and for any joint system where each joint has its own direction.This is an application of polygon cutting code DC of 3D DDA.
Polygon (computer graphics)
Cite
Citations (43)