lr-por / cl-conllu Goto Github PK
View Code? Open in Web Editor NEWtool for working with conllu files in CL
License: Apache License 2.0
tool for working with conllu files in CL
License: Apache License 2.0
A udapi-python ainda tem um modulo para gerar arvores em LaTeX, tikz.py . Saidas https://pt.wikipedia.org/wiki/SVG ou https://www.graphviz.org também seriam interessantes.
To reproduce:
(write-conllu (read-conllu #p"test") #p"test.out")
with the attached file. test.out
will be empty.
Functions for inserting and removing tokens from sentences are needed. It's necessary to not only to insert [remove] corresponding rows, but also re-enumerate ids (and references, at the head field) in other rows.
The following sentence breaks READ-CONLLU:
# text = #
# sent_id = 2016-06-22-F-zone-Geology-for-Rock-Mechanics-draft.pptx.potx34
1 # # NOUN NN Number=Sing 0 root _ TokenRange=1479:1480
with the following backtrace:
0: (SB-KERNEL:VECTOR-SUBSEQ* #<unavailable argument> #<unavailable argument> #<unavailable argument>)
1: (CL-CONLLU::COLLECT-META ("# text = #" "# sent_id = 2016-06-22-F-zone-Geology-for-Rock-Mechanics-draft.pptx.potx34"))
2: (MAKE-SENTENCE 302 ("# text = #" "# sent_id = 2016-06-22-F-zone-Geology-for-Rock-Mechanics-draft.pptx.potx34" "1 # # NOUN NN Number=Sing 0 root _ TokenRange=1479:1480") #<FUNCTION CL-CONLLU::COLLECT-M..
read/write json-ld from the internal sentence structure.
No arquivo projective.lisp temos uma "tradução" para lisp da função validate_projective_punctuation feita em python que está no repositório tools do UD.
https://github.com/UniversalDependencies/tools/blob/master/validate.py
A ideia é usar a função para verificar se a não projetividade de uma sentença é causada por uma pontuação. O problema é que mesmo para as sentenças que foram dadas válidas pelo código em python.
Os reports podem ser encontrados aqui:
https://github.com/UniversalDependencies/UD_Portuguese-Bosque/tree/workbench/reports/validation
a função em lisp retorna que há não projetividade causada pela pontuação.
In http://universaldependencies.org/format.html, few information is provided about the comments and metadata:
Lines starting with the # character and preceding a sentence are considered as carrying comments or metadata relevant to the following sentence.
We need to handle both cases:
# value
# key = value
I get this:
; in: DEFUN DEEP-AUX
; (CL-CONLLU::DEEP-AUX CL-CONLLU::CHILD CL-CONLLU:SENTENCE CL-CONLLU::FN-KEY)
;
; caught WARNING:
; The function was called with three arguments, but wants exactly two.
I don't see a need to roll our own edit distance function. There are plenty of libraries with a lot of different variants already available:
CL-CONLLU> (conllu.draw:tree-sentence (sentence-by-id "CF155-1" "documents/CF0155.conllu"))
─┮
│ ╭─╼ Esses det
│ ╭─┾ núcleos nsubj:pass
│ │ ├─╼ coloniais amod
│ │ │ ╭─╼ , punct
│ │ │ │ ╭─╼ entre case
│ │ ├ │─┶ eles nmod
│ │ ╰─┾ o appos
│ │ │ ╭─╼ de case
│ │ │ ├─╼ o det
│ │ ├─┾ Vale nmod
│ │ │ │ ╭─╼ de case
│ │ │ │ ├─╼ o det
│ │ │ ╰─┶ Ribeira nmod
│ │ ╰─╼ , punct
│ ├─╼ foram aux:pass
╰─┾ ocupados root
│ ╭─╼ a case
├─┾ partir obl
│ │ ╭─╼ de case
│ ╰─┾ terras nmod
│ ╰─┮ vendidas acl
│ │ ╭─╼ por case
│ │ ├─╼ o det
│ ╰─┶ governo obl:agent
╰─╼ . punct
data
# text = Esses núcleos coloniais, entre eles o do Vale do Ribeira, foram ocupados a partir de terras vendidas pelo governo.
# source = CETENFolha n=155 cad=Informática sec=com sem=94b
# sent_id = CF155-1
# id = 656
1 Esses esse DET <dem>|DET|M|P|@>N Gender=Masc|Number=Plur|PronType=Dem 2 det _ _
2 núcleos núcleo NOUN <np-def>|N|M|P|@SUBJ> Gender=Masc|Number=Plur 16 nsubj:pass _ _
3 coloniais colonial ADJ ADJ|M|P|@N< Gender=Masc|Number=Plur 2 amod _ ChangedBy=Issue165|SpaceAfter=No
4 , , PUNCT PU|@PU _ 7 punct _ _
5 entre entre ADP PRP|@<ADVL _ 6 case _ _
6 eles eles PRON PERS|M|3P|NOM/PIV|@P< Gender=Masc|Number=Plur|Person=3|PronType=Prs 2 nmod _ _
7 o o DET <dem>|DET|M|S|@N<PRED Gender=Masc|Number=Sing|PronType=Dem 2 appos _ _
8-9 do _ _ _ _ _ _ _ _
8 de de ADP <sam->|PRP|@N< _ 10 case _ _
9 o o DET <-sam>|<artd>|ART|M|S|@>N Definite=Def|Gender=Masc|Number=Sing|PronType=Art 10 det _ _
10 Vale Vale PROPN PROP|M|S|@P< Gender=Masc|Number=Sing 7 nmod _ MWE=Vale_do_Ribeira
11-12 do _ _ _ _ _ _ _ _
11 de de ADP <sam->|PRP|@N< _ 13 case _ _
12 o o DET <artd>|<-sam>|DET|M|S|@>N Definite=Def|Gender=Masc|Number=Sing|PronType=Art 13 det _ _
13 Ribeira Ribeira PROPN PROP|@P< Number=Sing 10 nmod _ ChangedBy=Issue165|SpaceAfter=No
14 , , PUNCT PU|@PU _ 7 punct _ _
15 foram ser AUX <aux>|V|PS/MQP|3P|IND|@FS-STA Mood=Ind|Number=Plur|Person=3|VerbForm=Fin 16 aux:pass _ _
16 ocupados ocupar VERB <pass>|<mv>|V|PCP|M|P|@ICL-AUX< Gender=Masc|Number=Plur|VerbForm=Part|Voice=Pass 0 root _ _
17 a a ADP PRP|@<ADVL _ 18 case _ MWE=a_partir_de
18 partir partir NOUN N|@P< _ 16 obl _ _
19 de de ADP PRP|@N< _ 20 case _ _
20 terras terra NOUN <np-idf>|N|F|P|@P< Gender=Fem|Number=Plur 18 nmod _ _
21 vendidas vender VERB <mv>|V|PCP|F|P|@ICL-N< Gender=Fem|Number=Plur|VerbForm=Part|Voice=Pass 20 acl _ _
22-23 pelo _ _ _ _ _ _ _ _
22 por por ADP PRP|@PASS _ 24 case _ _
23 o o DET <-sam>|<artd>|ART|M|S|@>N Definite=Def|Gender=Masc|Number=Sing|PronType=Art 24 det _ _
24 governo governo NOUN <np-def>|N|M|S|@P< Gender=Masc|Number=Sing 21 obl:agent _ ChangedBy=Issue165|SpaceAfter=No
25 . . PUNCT PU|@PU _ 16 punct _ _
We can now export CoNLL-U data a RDF. It'd be interesting to be able to read it in this format as well. We expected that wilbur would offer RDF parsing tools.
instead of lazy-stream-reader calling read-stream, it should be the other way around. The primitive function should consume a sentence from an open stream. all other functions should be on top of that.
We'll need functions for manipulating CoNLL-U files in order to extract semantic tags to some kind of RDF structure. These tags are returned, for instance, from Palavras or as a result of rule applications. These semantic tags will be at the MISC field.
After including Wilbur in cl-conllu.asd, it conflicted with rules.lisp, probably due to changes in the readtable. Replacing some symbols' names fixed the problem, but a better solution should be found.
Is it possible to isolate these changes to the readtable? (or perhaps even avoid them altogether, as we currently don't use them in rdf-wilbur.lisp)
In 4dca2b I solved the merge conflict and tested the corte-e-costura function.
The test was done using this conllu file and these rules:
(=> ((?a (= upostag "ADV") (= lemma "além"))
*
(?b (= lemma "ter")))
((?b (+ lemma "============"))))
(=> (?a (= lemma "liberdade") (= upostag "NOUN"))
(?a (! upostag "===========")))
(=> ((?a (= lemma "trabalho"))
?
(?b (= lemma "passar"))
?
(?c (= lemma "em")))
((?a (! lemma "============"))))
cl-conllu
fails to install on a pristine system since wilbur
is not a quicklisp package.
* (ql:quickload :wilbur)
debugger invoked on a QUICKLISP-CLIENT:SYSTEM-NOT-FOUND in thread
#<THREAD "main thread" RUNNING {1001950083}>:
System "wilbur" not found
not using UD rels and postags but is should be an conllu valid file:
# text = In contrast to most other salt basins, internal layering makes visible the high degree of internal deformation found in the Santos Basin.
# sent_id = 0
2 In_contrast_to in_contrast_to prep _ prep 10 vprep _ TokenRange=0:7|compslots=psubj:10,objprep:7
4 most most qual _ qual|superl 5 adjpre _ TokenRange=3:4
5 other other adj _ adj|superl|novadj|badcoordadj 7 nadj _ TokenRange=3:5|compslots=asubj:7
6 salt salt noun _ noun|cn|sg|massn|sbst|ent 7 nnoun _ TokenRange=5:6|compslots=u
7 basins basin noun _ noun|cn|pl|physobj|artf|inst|container|ent 2 objprep _ TokenRange=3:7|compslots=u
8 internal internal adj _ adj 9 nadj _ TokenRange=7:8|compslots=asubj:9,u
9 layering layering noun _ noun|cn|sg 10 subj _ TokenRange=7:9|compslots=u,u
10 makes make verb _ verb|vfin|vpres|sg|vsubj|badnen|vchng 0 top _ TokenRange=0:22|compslots=subj:9,obj:14,comp:11
11 visible visible adj _ adj 10 comp _ TokenRange=10:11|compslots=asubj:14,u,u
12 the the det _ det|sg|def|the|ingdet 14 ndet _ TokenRange=11:12
13 high high adj _ adj|erest|nqual|lmeasadj 14 nadj _ TokenRange=12:13|compslots=asubj:14,u
14 degree degree noun _ noun|cn|sg|locnoun|meas|lmeas|abst|property|massn|state 10 obj _ TokenRange=11:22|compslots=u,nobj:15,u
15 of of prep _ prep|pprefn|nonlocp 14 nobj _ TokenRange=14:17|compslots=psubj:14,objprep:17
16 internal internal adj _ adj 17 nadj _ TokenRange=15:16|compslots=asubj:17,u
17 deformation deformation noun _ noun|cn|sg|evnt|chng|hapning 15 objprep _ TokenRange=15:17|compslots=u,u,u
18 found find verb _ verb|ven|vpass|passen|sta|noptlo 14 nnfvp _ TokenRange=17:22|compslots=u,obj:14,comp:19
19 in in prep _ prep|staticp 18 comp _ TokenRange=18:22|compslots=psubj:18,objprep:22
20 the the det _ det|sg|pl|def|the|ingdet 22 ndet _ TokenRange=19:20
22 Santos_Basin Santos_Basin noun _ noun|propn|sg|pl|sgpl|glom|physobj|artf|inst|notfnd|container|ent 19 objprep _ TokenRange=19:22
if we set a token field to the empty string, it will be serialized as an empty string, which is an invalid format. It should be serialized as _
.
We need better error handling and messages.
In 0de6034 I added some files for test. Also in test.lisp we have some initial tests. We need more tests.
We need a good library for tests.
better integration with https://github.com/odanoburu/conllu-mode by @odanoburu
uma ideia que talvez melhore o seu work flow é ter o relatório num formato que o https://www.gnu.org/software/emacs/manual/html_node/emacs/Compilation-Mode.html entenda, aí vc pode clicar nos erros e já abrir a sentença. (vc teria de colocar o nome do arquivo e linha/coluna, acho)
draw need to call cl-conllu passing the sentence in the STDIN.
As it stands now, atoms for sentences are ctestset_scf790_2
(context testset, sentence cf790_2), while word atoms are like ctestset_i25
(context testset, word 25). For instance, we have:
?- nlp_sentence(S), nlp_dependency(S,Y,Z,W).
S = ctestset_scf790_2,
Y = ctestset_i25,
Z = ctestset_i9,
W = punct
However, for analysing multiple sentences this isn't so great, as the 1st word from different sentences of same context are both called cCONTEXT_i1
.
Perhaps we should add a sentence identifier on each word atom as well.
http://universaldependencies.org/format.html
The read and write functions still do not deal with the multiword lines.
https://github.com/udapi/udapi-python/blob/master/udapi/block/write/textmodetrees.py#L20
We need some similar feature for printing a sentence as a tree. This trees are very useful for visualize the data:
─┮
╰─┮ Gosto VERB root
│ ╭─╼ de ADP mark
├─┾ levar VERB xcomp
│ │ ╭─╼ a ADP case
│ ├─┶ sério NOUN xcomp
│ │ ╭─╼ o DET det
│ │ ├─╼ meu DET det
│ ╰─┾ papel NOUN obj
│ │ ╭─╼ de ADP case
│ ╰─┾ consultor NOUN nmod
│ ╰─╼ encartado VERB acl
╰─╼ . PUNCT punct
alternatives:
convert conllu to tex and compile it
udapy write.Tikz attributes=form,lemma,upos < my.conllu > my.tex
If needed I can add more features to
https://github.com/udapi/udapi-python/blob/master/udapi/block/write/tikz.py
e.g. printing multiword tokens and some default colors.
Of course, for camera-ready pictures a bit of manual fine-tuning of the layout will be needed.
You can try also
udapy write.TextModeTrees color=1 < my.conllu | less -R
output above.
https://github.com/udapi/udapi-python/blob/master/udapi/block/write/textmodetrees.py
either use in verbatim without colors, or subclass so it generates TeX syntax for colors (and texttt/verb), similarly as in
https://github.com/udapi/udapi-python/blob/master/udapi/block/write/textmodetreeshtml.py
https://github.com/udapi/udapi-python/blob/master/udapi/block/write/html.py
for html+javascript output like
http://ufallab.ms.mff.cuni.cz/~popel/czeng1.6-sample.html
There is a button for SVG export and you can use
inkscape -D -z --file=image.svg --export-pdf=image.pdf --export-latex
to export it to pdf and tex:
\begin{figure}
\centering
\def\svgwidth{\columnwidth}
\input{image.pdf_tex}
\end{figure}
Can I outout to LaTeX the
second command :udapy write.TextModeTrees color=1 < my.conllu | less -R
Yes, but without the colors:
echo '\begin{verbatim}' > my.tex
udapy write.TextModeTrees < my.conllu >> my.tex
echo '\end{verbatim}' >> my.tex
and then use
\input{my.tex}
It would not be difficult to write a subclass of write.TextModeTrees
which would use some LaTeX markup like \lemma{I}, \upos{PRON}
instead of the ANSI color codes. So then you could define the colors&style
\def\lemma#1{\textcolor{red}{#1}}
If you are interested, I can implement it.
what I really missing is a simple way to display a fragment of a sentence
Now, I've added a Udapi block which allows to delete all nodes in a document
except for the subtrees matching a given condition, e.g.
udapy -s util.Filter subtree='node.upos == "NOUN"' < in.conllu > filtered.conllu
will print only noun phrases.
So you can use
udapy util.Filter subtree='node.form == "dog"' write.TextModeTrees < in.conllu
to get the subtree(s) headed by word "dog", or
udapy util.Filter subtree='node.ord == 2 and node.root.address() == "3"' write.TextModeTrees < in.conllu
to get the subtree headed by the second word in tree with sent_id = 3.
Yet another alternative to Tikz, Html and TextModeTrees would be to
use paste the CoNLL-U to the online Brat rendered
(e.g. click "edit" here http://universaldependencies.org/sandbox.html#pirate-example).
But then you would need to zoom, take a screenshot and include it as bitmap (png) into LaTeX,
which is not optimal.
If needed I can implement write.Sdparse which would print something like
Dogs run
nsubj(run-2, Dogs-1)
which would allow easier manual editing than the CoNLL-U format.
In the https://github.com/own-pt/cl-conllu/blob/master/read-write.lisp we need a function to read the PALAVRAS output like https://github.com/cpdoc/dhbb/blob/master/pal/100.dep to CoNLL-U format.
In #15 we added the vertical tree visualisation. It 'd be good to allow the user to specify what info from tokens he wants in the tree.
The current implementation of queries do not compile them, the queries are evaluated for EVERY sentence. A compilation into a lambda expression would improve performance.
currently we save in conllu
, rdf
, and prolog
, through different and inconsistent methods. we should unify them under a single and uniform API.
To add a function that verifies if a dependency parse of a sentence is projective or not.
I've done already two commits about it: 79bcc35 and 7c6f700, but testing (and possibly improving the code) is still necessary.
According to http://universaldependencies.org/format.html metadata needs to be separate between a =
between key and value.
We need to:
This way we can easily convert existing CONLL-U files.
we need also to represent features / MISC and the complete info in the CoNLL-U file.
# text = O compromisso com a precisão (e, por extensão, com o leitor) vale menos do que a torcida da imprensa nessas horas.
# sent_id = CF670-8
# source = CETENFolha n=670 cad=Brasil sec=pol sem=94b &D
# id = 2810
1 O o DET <artd>|ART|M|S|@>N Definite=Def|Gender=Masc|Number=Sing|PronType=Art 2 det _ _
2 compromisso compromisso NOUN <np-def>|N|M|S|@SUBJ> Gender=Masc|Number=Sing 16 nsubj _ _
3 com com ADP <first-cjt>|PRP|@N< _ 5 case _ _
4 a o DET <artd>|ART|F|S|@>N Definite=Def|Gender=Fem|Number=Sing|PronType=Art 5 det _ _
5 precisão precisão NOUN <first-cjt>|<np-def>|N|F|S|@P< Gender=Fem|Number=Sing 2 nmod _ _
6 ( ( PUNCT PU|@PU _ 10 punct _ ChangedBy=Issue165|SpaceAfter=No
7 e e CCONJ <co-postnom>|KC|@CO _ 14 cc _ ChangedBy=Issue165|SpaceAfter=No|d2d:#106
8 , , PUNCT PU|@PU _ 9 punct _ d2d:#106
9 por por ADP PRP|@<ADVL _ 10 case _ _
10 extensão extensão NOUN <np-idf>|N|F|S|@P< Gender=Fem|Number=Sing 5 nmod _ ChangedBy=Issue165|SpaceAfter=No
11 , , PUNCT PU|@PU _ 9 punct _ _
12 com com ADP <cjt>|PRP|@N< _ 14 case _ _
13 o o DET <artd>|ART|M|S|@>N Definite=Def|Gender=Masc|Number=Sing|PronType=Art 14 det _ _
14 leitor leitor NOUN <np-def>|N|M|S|@P< Gender=Masc|Number=Sing 5 conj _ ChangedBy=Issue165|SpaceAfter=No
15 ) ) PUNCT PU|@PU _ 10 punct _ _
16 vale valer VERB <mv>|V|PR|3S|IND|@FS-STA Mood=Ind|Number=Sing|Person=3|Tense=Pres|VerbForm=Fin 0 root _ _
17 menos pouco ADV <quant>|<KOMP>|<COMP>|ADV|@<ADVL _ 16 advmod _ _
18-19 do _ _ _ _ _ _ _ _
18 de de ADP <sam->|PRP|@COM _ 22 case _ MWE=do_que
19 o o PRON <dem>|<-sam>|DET|M|S|@P< Gender=Masc|Number=Sing|PronType=Dem 18 fixed _ _
20 que que PRON <rel>|INDP|M|S|@N< Gender=Masc|Number=Sing|PronType=Rel 18 fixed _ _
21 a o DET <artd>|ART|F|S|@>N Definite=Def|Gender=Fem|Number=Sing|PronType=Art 22 det _ _
22 torcida torcida NOUN <first-cjt>|<np-def>|N|F|S|@KOMP< Gender=Fem|Number=Sing 17 obl _ _
23-24 da _ _ _ _ _ _ _ _
23 de de ADP <sam->|PRP|@N< _ 25 case _ _
24 a o DET <-sam>|<artd>|ART|F|S|@>N Definite=Def|Gender=Fem|Number=Sing|PronType=Art 25 det _ _
25 imprensa imprensa NOUN <np-def>|N|F|S|@P< Gender=Fem|Number=Sing 22 nmod _ _
26-27 nessas _ _ _ _ _ _ _ _
26 em em ADP <sam->|PRP|@<ADVL _ 28 case _ _
27 essas esse DET <-sam>|<dem>|DET|F|P|@>N Gender=Fem|Number=Plur|PronType=Dem 28 det _ _
28 horas hora NOUN <np-def>|N|F|P|@P< Gender=Fem|Number=Plur 16 obl _ ChangedBy=Issue137|ChangedBy=Issue165|SpaceAfter=No
29 . . PUNCT PU|@PU _ 16 punct _ _
Fix bug at
https://github.com/own-pt/cl-conllu/blob/master/data.lisp#L63-L75
(defun sentence->deep (sentence &key fn-key)
(labels ((ensure-list (key)
(if (symbolp key) (list fn-key) key)))
(if (functionp fn-key)
(deep-aux (sentence-root sentence) sentence fn-key)
(if (or (symbolp fn-key)
(listp fn-key))
(deep-aux (sentence-root sentence) sentence
(lambda (tk)
(let ((out (loop for k in (ensure-list fn-key)
collect (slot-value tk k))))
(if (and (listp out) (= 1 (length out)))
(car out) out))))))))
(insert bug description or example of error?)
t
or nil
.Also: ideally it should provide the data in a structured way so that visual editors can use it to highlight the errors in the original file, not just display the error message in the screen. This is intentionally vague -- we need to define this better, of course.
The idea is two be able to edit the sentences via a local web
interface. Working in progress.
Lisp Backend:
Javascript and CSS:
Table editor ideas:
https://editor.datatables.net/examples/inline-editing/simple
https://codepen.io/ashblue/pen/mCtuA
http://markcell.github.io/jquery-tabledit/#home
http://www.jtable.org
It's useful to have functions for evaluating parsers (UAS, LAS, precision, ecall, confusion matrix, etc), testing results of sentences against a golden dataset. I'll use them myself this week.
I'm starting the eval-functions branch in order to develop this and already did the first commit: 439d192.
We are using the term token when we should be using word. This would require rename the token class and some names of functions and variables
http://universaldependencies.org/u/overview/tokenization.html
A number of useful methods written for the scripts to fix the issues in Bosque-UD can be imported to this library. Things like isolated?
, add-feature
, remove-feature
, search-features
, etc. They should be adapted for the general case and added here.
Optionally (and maybe this is a larger issue) we also started a very simple query language, implemented by the method t?
in the scripts that can be finalized and imported here as well. This one, however, needs a better implementation as it is extremely simple, just enough functionality to handle the requirements of those scripts.
Recommend looking at the latest scripts (by date) to see the more fully featured versions of those methods.
Given that we have already the classes sentence and tokens, can we make a query interface?
Last link, query example:
#x:[pos="VERB" & word & lemma="ficar"] . #y:[pos="VERB" & word & lemma]
another alternative is to have a backend (ES, Solr, triple store) and a mapping from a expression lang to the backend query language.
http://wesearch.delph-in.net/deepbank/search.jsp
another option, convert to prolog and use any of CL implementations of PROLOG:
A query would be something like de expression below that could filter sentences from a list of sentences:
(lemma tk1 "fazer") and (pos tk1 VERB) and (pos tk2 VERB) and (aux tk1 tk2)
Likely the presence of SUBJ_INDEF
on the misc
field, but it is expecting that values in that field are always of the form Key=Value
, which is not the case.
debugger invoked on a SB-KERNEL::ARG-COUNT-ERROR in thread
#<THREAD "main thread" RUNNING {1001BB6A83}>:
error while parsing arguments to DESTRUCTURING-BIND:
too few elements in
("SUBJ_INDEF")
to satisfy lambda list
(NAME VALUE):
exactly 2 expected, but got 1
https://corpling.uis.georgetown.edu/depedit/
good ideas
has support for empty tokens been added yet? I'm mirroring this issue from hs-conllu
so we can fix this in both libraries. a CoNLL-U file that can be used for testing is this one.
(I'm asking and not testing myself because I'm getting this error when loading the library:
COMPILE-FILE-ERROR while
compiling #<CL-SOURCE-FILE "iterate" "iterate">
[Condition of type UIOP/LISP-BUILD:COMPILE-FILE-ERROR]
)
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google ❤️ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.