Semantic graph parser based on Categorial grammars

Overview

Lambekseq

semgraph

"Everyone who failed Greek or Latin hates it."


This package is for proving theorems in Categorial grammars (CG) and constructing semantic graphs, i.e., semgraphs on top of that.

Three CG calculuses are supported here (see below). A "proof" is simply a set of atom links, abstracting away from derivaiton details.

Requirements

Add the path to the package to PYTHONPATH. None of the below packages is needed to use the theorem proving facility.

Semantic graphs derive from digraph:

For graph visualization we use

Background

This package is used for the author's PhD thesis in progress.

Categorial grammars:

Semantic graphs:

Theorem Proving

To prove a theorem, use atomlink module. For example, using Lambek Calculus to prove np np\s -> s.

>>> import lambekseq.atomlink as al

>>> con, *pres = 's np np\\s'.split()
>>> con, pres, parser, _ = al.searchLinks(al.LambekProof, con, pres)
>>> al.printLinks(con, pres, parser)

This outputs

----------
s_0 <= np_1 np_2\s_3

(np_1, np_2), (s_0, s_3)

Total: 1

You can run atomlink in command line. The following finds proofs for the theorems in input, using abbreviation definitions in abbr.json and Contintuized CCG.

$ python atomlink.py -i input -a abbr.json -c ccg --earlyCollapse

Theorem s qp vp/s qp vp (the first item is the conclusion, the rest the premises) is thus proved as follows:

<class 'lambekseq.cntccg.Cntccg'>
----------
s_0 <= (s_1^np_2)!s_3 (np_4\s_5)/s_6 (s_7^np_8)!s_9 np_10\s_11

(np_10, np_8), (np_2, np_4), (s_0, s_3), (s_1, s_5), (s_11, s_7), (s_6, s_9)

Total: 1

When using Lambek/Displacement/CCG calculus, you can also inspect the proof tree that yields atom links:

>>> con, *pres = 's', 'np', '(np\\s)/np', 'np'
>>> con, pres, parser, _ = al.searchLinks(al.LambekProof, con, pres)
>>> parser.buildTree()
>>> parser.printTree()
(np_1, np_2), (np_4, np_5), (s_0, s_3)
........ s_3 -> s_0
........ np_1 -> np_2
.... np_1 np_2\s_3 -> s_0
.... np_5 -> np_4
 np_1 (np_2\s_3)/np_4 np_5 -> s_0

You can export the tree to Bussproofs code for Latex display:

bussproof

>>> print(parser.bussproof)
...
\begin{prooftree}
\EnableBpAbbreviations
        \AXC{s$_{3}$ $\to$ s$_{0}$}
        \AXC{np$_{1}$ $\to$ np$_{2}$}
    \BIC{np$_{1}$\enskip{}np$_{2}$\textbackslash s$_{3}$ $\to$ s$_{0}$}
    \AXC{np$_{5}$ $\to$ np$_{4}$}
\BIC{np$_{1}$\enskip{}(np$_{2}$\textbackslash s$_{3}$)/np$_{4}$\enskip{}np$_{5}$ $\to$ s$_{0}$}
\end{prooftree}

Run python atomlink.py --help for details.

Semantic Parsing

Use semcomp module for semantic parsing. You need to define graph schemata for parts of speech as in schema.json.

>>> from lambekseq.semcomp import SemComp
>>> SemComp.load_lexicon(abbr_path='abbr.json',
                         vocab_path='schema.json')
>>> ex = 'a boy walked a dog'
>>> pos = 'ind n vt ind n'
>>> sc = SemComp(zip(ex.split(), pos.split()), calc='dsp')
>>> sc.unify('s')

Use graphviz's Source to display the semgraphs constructed from the input:

>>> from graphviz import Source
>>> Source(sc.semantics[0].dot_styled)

This outputs
a boy walked a dog

You can inspect the syntax behind this parse:

>>> sc.syntax[0].insight.con, sc.syntax[0].insight.pres
('s_0', ['np_1/n_2', 'n_3', '(np_4\\s_5)/np_6', 'np_7/n_8', 'n_9'])

>>> sc.syntax[0].links
['(n_2, n_3)', '(n_8, n_9)', '(np_1, np_4)', '(np_6, np_7)', '(s_0, s_5)']

See demo/demo.ipynb for more examples.

You can export semgraphs to tikz code that can be visually edited by TikZit.

a boy walked a dog

>>> print(sc.semantics[0].tikz)
\begin{tikzpicture}
\begin{pgfonlayer}{nodelayer}
        \node [style=node] (i1) at (-1.88,2.13) {};
        \node [style=none] (g2u0) at (-2.99,3.07) {};
        \node [style=node] (i0) at (0.99,-2.68) {};
        \node [style=none] (g5u0) at (1.09,-4.13) {};
        \node [style=node] (g3a0) at (0.74,0.43) {};
        \node [style=none] (g3u0) at (2.05,1.19) {};
        \node [style=none] (0) at (-3.04,2.89) {boy};
        \node [style=none] (1) at (0.61,-4.00) {dog};
        \node [style=none] (2) at (-0.66,0.72) {ag};
        \node [style=none] (3) at (0.63,-0.77) {th};
        \node [style=none] (4) at (2.42,1.09) {walked};
\end{pgfonlayer}
\begin{pgfonlayer}{edgelayer}
        \draw [style=arrow] (i1) to (g2u0.center);
        \draw [style=arrow] (i0) to (g5u0.center);
        \draw [style=arrow] (g3a0) to (i1);
        \draw [style=arrow] (g3a0) to (i0);
        \draw [style=arrow] (g3a0) to (g3u0.center);
\end{pgfonlayer}
\end{tikzpicture}
A PyTorch implementation of NeRF (Neural Radiance Fields) that reproduces the results.

NeRF-pytorch NeRF (Neural Radiance Fields) is a method that achieves state-of-the-art results for synthesizing novel views of complex scenes. Here are

Yen-Chen Lin 3.2k Jan 08, 2023
Model serving at scale

Run inference at scale Cortex is an open source platform for large-scale machine learning inference workloads. Workloads Realtime APIs - respond to pr

Cortex Labs 7.9k Jan 06, 2023
Source code for Zalo AI 2021 submission

zalo_ltr_2021 Source code for Zalo AI 2021 submission Solution: Pipeline We use the pipepline in the picture below: Our pipeline is combination of BM2

128 Dec 27, 2022
E2EDNA2 - An automated pipeline for simulation of DNA aptamers complexed with small molecules and short peptides

E2EDNA2 - An automated pipeline for simulation of DNA aptamers complexed with small molecules and short peptides

11 Nov 08, 2022
CFNet: Cascade and Fused Cost Volume for Robust Stereo Matching(CVPR2021)

CFNet(CVPR 2021) This is the implementation of the paper CFNet: Cascade and Fused Cost Volume for Robust Stereo Matching, CVPR 2021, Zhelun Shen, Yuch

106 Dec 28, 2022
Lightweight library to build and train neural networks in Theano

Lasagne Lasagne is a lightweight library to build and train neural networks in Theano. Its main features are: Supports feed-forward networks such as C

Lasagne 3.8k Dec 29, 2022
Official pytorch implementation of paper "Inception Convolution with Efficient Dilation Search" (CVPR 2021 Oral).

IC-Conv This repository is an official implementation of the paper Inception Convolution with Efficient Dilation Search. Getting Started Download Imag

Jie Liu 111 Dec 31, 2022
End-to-end beat and downbeat tracking in the time domain.

WaveBeat End-to-end beat and downbeat tracking in the time domain. | Paper | Code | Video | Slides | Setup First clone the repo. git clone https://git

Christian J. Steinmetz 60 Dec 24, 2022
Ensembling Off-the-shelf Models for GAN Training

Vision-aided GAN video (3m) | website | paper Can the collective knowledge from a large bank of pretrained vision models be leveraged to improve GAN t

345 Dec 28, 2022
Topic Discovery via Latent Space Clustering of Pretrained Language Model Representations

TopClus The source code used for Topic Discovery via Latent Space Clustering of Pretrained Language Model Representations, published in WWW 2022. Requ

Yu Meng 63 Dec 18, 2022
Generalized Data Weighting via Class-level Gradient Manipulation

Generalized Data Weighting via Class-level Gradient Manipulation This repository is the official implementation of Generalized Data Weighting via Clas

18 Nov 12, 2022
EMNLP 2021: Single-dataset Experts for Multi-dataset Question-Answering

MADE (Multi-Adapter Dataset Experts) This repository contains the implementation of MADE (Multi-adapter dataset experts), which is described in the pa

Princeton Natural Language Processing 68 Jul 18, 2022
Keras implementation of "One pixel attack for fooling deep neural networks" using differential evolution on Cifar10 and ImageNet

One Pixel Attack How simple is it to cause a deep neural network to misclassify an image if an attacker is only allowed to modify the color of one pix

Dan Kondratyuk 1.2k Dec 26, 2022
Talk covering the features of skorch

Skorch Talk Skorch - A Union of Scikit-learn and PyTorch Presentation The slides can be downloaded at: download link. Google Colab Part One - MNIST Pa

Thomas J. Fan 3 Oct 20, 2020
[ICLR 2021 Spotlight Oral] "Undistillable: Making A Nasty Teacher That CANNOT teach students", Haoyu Ma, Tianlong Chen, Ting-Kuei Hu, Chenyu You, Xiaohui Xie, Zhangyang Wang

Undistillable: Making A Nasty Teacher That CANNOT teach students "Undistillable: Making A Nasty Teacher That CANNOT teach students" Haoyu Ma, Tianlong

VITA 71 Dec 28, 2022
object detection; robust detection; ACM MM21 grand challenge; Security AI Challenger Phase VII

赛题背景 在商品知识产权领域,知识产权体现为在线商品的设计和品牌。不幸的是,在每一天,存在着非法商户通过一些对抗手段干扰商标识别来逃避侵权,这带来了很高的知识产权风险和财务损失。为了促进先进的多媒体人工智能技术的发展,以保护企业来之不易的创作和想法免受恶意使用和剽窃,因此提出了鲁棒性标识检测挑战赛

65 Dec 22, 2022
Hso-groupie - A pwnable challenge in Real World CTF 4th

Hso-groupie - A pwnable challenge in Real World CTF 4th

Riatre Foo 42 Dec 05, 2022
Codes for CIKM'21 paper 'Self-Supervised Graph Co-Training for Session-based Recommendation'.

COTREC Codes for CIKM'21 paper 'Self-Supervised Graph Co-Training for Session-based Recommendation'. Requirements: Python 3.7, Pytorch 1.6.0 Best Hype

Xin Xia 42 Dec 09, 2022
CR-FIQA: Face Image Quality Assessment by Learning Sample Relative Classifiability

This is the official repository of the paper: CR-FIQA: Face Image Quality Assessment by Learning Sample Relative Classifiability A private copy of the

Fadi Boutros 33 Dec 31, 2022
[AAAI 2022] Sparse Structure Learning via Graph Neural Networks for Inductive Document Classification

Sparse Structure Learning via Graph Neural Networks for inductive document classification Make graph dataset create co-occurrence graph for datasets.

16 Dec 22, 2022