COURSEWORK 2 Natural Language Processing
CS-265/CSC325 Ar@ficial Intelligence
Released: 22 March 2024
Due: Thursday 29 April 2024, 11am. This is a hard deadline and set in rela-on to other submission
deadlines.
Read and think through the whole coursework before star5ng to program. Review the NLP lectures and
lab. See it as engineering a language mechanism, which you can experiment with and develop. The focus
is on how features are used at different points in the grammar and lexicon to control well-formedness.
Follow the instruc5ons fully and accurately. Marks are taken off for wrong or missing answers.
Address the following in the coursework
1. Write a DCG which parses input sentences and outputs a parse.
2. The DCG must parse sentences with the following features. The features a, b, c, d, and j are
related to the lab. The others - e, g, h, k, and l - are addi5ons that interact with other parts of the
grammar and are explained further below.
a. Transi5ve and intransi5ve verbs
b. Common nouns
c. Determiners (e.g., ``a'', ``the'', and ``two'')
d. Subject/object singular/plural pronouns (e.g., ``he'', ``him'', ``I", ``we'')
e. Pronouns with gramma5cal person (e.g., ``I'', ``you'', ``she'')
f. Singular/plural nouns (e.g., chair/chairs)
g. Adjec5ves and adjec5val phrases
h. Preposi5ons and preposi5onal phrases
i. Subject-verb agreement for person, number, and animacy
j. Determiner-common noun number agreement
k. Animacy agreement between the subject and verb
l. Passive and ac5ve sentences
3. The DCG should separate grammar and lexicon. The lexicon must be included in your code and
include all the words and word forms in the lexicon below or those you should add to the
lexicon.
4. The sample outputs (below) should be carefully studied and emulated by the parser. It must be
emphasised that the outputs are phrase structures of the input sentence; it is not sufficient
just to recognise if a sentence is well-formed according to the grammar. The categories, e.g.,
nbar, jp, adj, n, as well as the parsing structure, e.g., nbar(jp(adj(tall), jp(adj(young), n(men)))),
should appear in the output. Output that misses categories and parsing structures will be
marked down. While there may be different ways to write the grammar, the input and output
must be precise and fixed, as the results will be automa5cally checked. If the output does not
match the intended output, you can be marked down on that output.
Demonstra@on of Work
To demonstrate that your code works as intended, your code should correctly give the parse trees for the
gramma5cal sentences in the list of test sentences (below) and fail for the ungramma5cal sentences. In
the list of test sentences below, an ungramma5cal sentence is indicated with a * next to it, for example,
"*the men sees the apples" is ungramma5cal. We are not concerned with capitalisa5on or punctua5on.
For each sentence in the list of test sentences, query your parser; if the sentence is ungramma5cal, the
output should be false/fails; if the sentence is gramma5cal, the outputs should be the correct parse. An
excep5on in your program means there is a problem and no marks are given.
Generally, if issues or problems arise, report these in a discussion sec5on.
Your grammar should at least parse and provide the phrase structure for every sentence in the test
sentences (below). As well, for evalua5on, there will be unseen sentences that your grammar should
parse and generate the phrase structure or fail to parse, given the instruc5ons and the lexicon.
There are extensions that you are to make to the lexicon - read the instruc5ons carefully all the way
through to fully understand what needs to be done.
Submission
To submit your coursework, it should be one file with your grammar. The file name should a Prolog file in
the form of:
YOUR-STUDENT-NUMBER_AI_NLP_2024.pl
The files will be submiced on TurnItIn (the link to be provided). The grammars will be automa5cally run
and unit tested; that is, we will run your grammar against all the seen test sentences as well as unseen
sentences that your grammar should parse or fail to parse given the instruc5ons and the lexicon. This
also means that if you have anything in your file which is not Prolog code (comments, discussion,
examples, etc), then it should appear commented out.
Discussion
If you are inclined to engage with further discussion, issues, work on other examples, make observa5ons,
add further extensions, or even other languages, you are welcome to share it in an email document sent
to me (Adam Wyner). There is no addi5onal mark per se. You will get remarks in return from the
lecturer. Make sure you indicate your student number and name on your document.
Marking Scheme
The overall mark for the coursework is 15 marks. 60% of these marks are for correct output of parsed
sentences (seen and unseen data) and 40% of these marks are for a well-formed grammar. Incorrect
output (wholly or par5ally) are deducted propor5onately. A well-formed grammar uses DCGs (not
difference lists), uses the full lexicon with addi5ons (see below), uses the indicated gramma5cal
categories and phrasal units, and reduces redundancy or complexity as much as is feasible (see below for
further notes).
Test Sentences
* means the grammar should fail on these sentences. Some of the sentences may seem gramma5cal
given other interpreta5ons, which is discussed further below. Accept the gramma5cality judgements
given here, though they are open to discussion. Below, there are listed sentences; further below, there
are addi5onal sample input/outputs.
1. the woman sees the apples
2. a woman knows him
3. *two woman hires a man
4. two women hire a man
5. she knows her
6. *she know the man
7. *us see the apple
8. we see the apple
9. i know a short man
10. *he hires they
11. two apples fall
12. the apple falls
13. the apples fall
14. i sleep
15. you sleep
16. she sleeps
17. *he sleep
18. *them sleep
19. *a men sleep
20. *the tall woman sees the red
21. the young tall man knows the old short woman
22. *a man tall knows the short woman
23. a man on a chair sees a woman in a room
24. *a man on a chair sees a woman a room in
25. the tall young woman in a room on the chair in a room in the room sees the red apples under
the chair
26. the man sleeps
27. *the room sleeps
28. *the apple sees the chair
29. *the rooms know the man
30. the apple falls
31. the man falls
32. the man breaks the chairs
33. the chairs are broken
34. the chairs are broken by the man
35. the chair is broken by the men
36. *the chair is broken by the apple
37. the room is hired by the man
38. *the chair broken by the man
39. a man is bicen by the dog
40. *a man is bicen the woman by the dog
41. *the chair falls by the man
42. *a man on a chair sees a woman by her
43. *a man on a chair sees a woman by she
44. *a man on a chair sees a woman by the woman
45. a man on a chair sees a woman on a chair
In addi5on to these 45 sentences, there 15 sample input and parses below that you should use to
develop and test your grammar.
The 60 sentences is the data set that you can use to develop and test your grammar. However, there will
be more in the unseen tes5ng data set, using the same lexicon with the same parameters of the
grammar as described below. If your grammar parses and generates the phrase structure for the seen
examples, it should, assuming you’ve designed the grammar well, also parse and generate the phrase
structure for the unseen sentences.
Modeling the Sample Inputs and Parses
Below, in different sec5ons, you will find sample model inputs and outputs. They highlight the categories
and structures that your grammar should recognise and output their structure. The output in the
examples should be carefully studied and emulated by the grammar. The categories (below), e.g., nbar,
jp, adj, n, byPass, and parsing structures (below), e.g., nbar(jp(adj(tall), jp(adj(young), n(men)))), should
appear correctly in the output. Output that misses categories and parsing structures will be marked
down. The examples also illustrate the predicate that can be called. It is essen5al that you use the form
of this predicate and that your grammar produce these outputs in these forms. Obviously, the task of
your grammar is to take the input and provide the output, so you know the intended target.
Comments and Tips
The focus of the grammar in the coursework is on how features are used to ``guide'' well-formedness of
parses.
Your grammar ought to provide only one parse for each input sentence. Check that it makes sense given
the specifica5on. If there is more than one parse, there is something to revise in your grammar.
In a long parse, you might see .... your parse. This means that the parse is very long and Prolog is
trunca5ng it. If you want to see the full parse, let the lecturer know and a predicate can be circulated on
Canvas.
The length of the parses should be propor5onal to the length of the input sentence. If you have very long
parses for a rela5vely short sentence, then something is wrong with your grammar.
The grammar you are wri5ng should recognize and output the parse of the relevant sentences (those in
the seen data and others unseen rela5ve to the lexicon, sample parses, and gramma5cal construc5ons)
and fail on others. If you generate more sentences or provide further examples for parsing, you will
quickly see that there are many odd or ungramma5cal sentences that this grammar recognises. You will
also see that some sentences can be gramma5cal and given a different interpreta5on of some of the
parts of speech, e.g., the by-phrase only appears in the passive in this grammar, but in the meaning of
``alongside'' would be gramma5cal. In this sense, a grammar is a theory which you can develop and
evaluate incrementally with respect to the data. Wri5ng a large scale grammar for a fragment of natural
language must take into account a range of proper5es, e.g. ordering of preposi5onal phrases, alterna5ve
interpreta5ons, seman5c restric5ons, seman5c representa5ons, pragma5cs, etc., which we are not
addressing in this coursework. Going ``hard core'' in the world of computa5onal linguis5c parsing and
seman5c representa5on means facing lots of hard, complex, and very interes5ng issues of natural
language.
During development, you can also visualise the parse trees in SWISH (and probably not in your local SWIProlog installa5on). The predicate for this will be circulated on Canvas.
In general, your grammar file should only include the grammar and lexicon and no further Prolog
direc5ves (those lines that start with :) or predicates not part of the grammar and lexicon, i.e., such as
you might use in SWISH for parse trees.
Notes on Gramma@cal Construc@ons
The notes here which extend the topics found in Lab 6, so take them together.
Nouns (common and pronoun) carry features. This is explicit on pronoun forms (e.g. he/him, I/we),
where have features such as number (singular/plural) and case (nomina5ve/accusa5ve). The
gramma5cal role (subject/object) is related to the case on nouns, where the cases are nomina5ve and
accusa5ve and align respec5vely with subject and object (broadly); that is, case forms of nouns indicate
the role the noun has with respect to the verb.
Phrases and Structure
It is reasonable to have more than one rule for similar phrasal categories, where there is a significant
reason to warrant them. For instance, we have transi5ve verbs (must have an object) and intransi5ve
verbs (cannot have an object); you might have two different VP rules to represent this. Preposi5onal
phrases are op5onal inside Noun Phrases; the byPass preposi5onal phrases are op5onal inside Verb
Phrases. You might have two different NPs for the former, and two different VPs for the lacer. We do not
assume that binary branching is necessary, so a phrase might have more than two cons5tuents (parts)
within it.
s(Tree, [the,man,sleeps], []).
s(np(det(the),nbar(n(man))),vp(v(sleeps)))
s(Tree, [the,woman,sees,the,apples], []).
s(np(det(the),nbar(n(woman))),vp(v(sees),np(det(the),nbar(n(apples)))))
s(Tree, [the,woman,sees,the,apples,in,the,room], []).
s(np(det(the),nbar(n(woman))),vp(v(sees),np(det(the),nbar(n(apples)),pp(prep(in),np(det(the),n
bar(n(room)))))))
Pronouns
Pronouns (e.g. he/him, I/we) have features such as number (singular/plural) and gramma5cal role
(subject/object). The gramma5cal role is related to the case on nouns, that is, forms of nouns that
indicate what role the noun has with respect to the verb. In English, there are three cases - Nomina5ve,
Accusa5ve, and Geni5ve. The last we ignore. Pronouns show this most clearly in English - ``she'' is a
pronoun in the nomina5ve form, while ``her'' is a pronoun in the accusa5ve form. When a pronoun is in
the subject posi5on, it must appear in the nomina5ve form; when a pronoun is in the object posi5on, it
must appear in the accusa5ve form. In addi5on, pronouns have features such as gramma5cal person,
e.g. first person ``i'', second person ``you'', third person ``she''. Gramma5cal person indicates a closer or
more distant rela5onship between the speaker of the sentence and other persons: "I see the apple"
represents the most personal statement (first person); ``You see the apple'' is between the speaker and a
person who is immediately present; and ``He sees the apple'' is the most distant, as it can relate to a
person who is not immediately present or somehow less ``relevant''.
The lexicon shows the features of pronouns number, case, gramma5cal person, and animacy (whether it
is a cogni5ve en5ty).
s(Tree,[she,knows,her],[]).
Tree = s(np(pro(she)), vp(v(knows), np(pro(her)))).
s(Tree,[her,knows,she],[]).
false.
Agreement
Pronouns and noun phrases show several features. A noun phrase (and pronouns in par5cular) must
agree with the verb in several ways, that is, the number, case, person, and animacy features of the noun
phrase must be compa5ble with those features of the verb - this is how the structure of the mechanism
'locks' together. While sentences have subject and object posi5ons in sentences, these are reflected in
the order of arguments rather than some addi5onal feature.
Structures for NPs with Adjec-ves and Preposi-onal Phrases
For our purposes, an adjec5ve such as ``tall'' describes a property of a common noun such a man. The
adjec5ve precedes the noun. For example: ``the tall man sees the woman'' is gramma5cal; ``the man tall
sees the woman'' is ungramma5cal. You can have any number of adjec5ves, for example: ``the tall tall
old man sees the woman''; ``the tall tall old old man sees the woman'', even if a bit odd, we'll accept as
gramma5cal.
For our purposes, a preposi5onal phrase modifies a noun without restric5on, and it is made up of a
preposi5on and a noun phrase. The preposi5on provides informa5on about the rela5ve loca5ons of the
nouns i.e., the noun that is modified and the noun within the preposi5onal phrase. The preposi5onal
phrase follows the noun that it modifies: ``the man in the room sees a woman on a chair''. We see them
as a rela5on between ``man'' and ``room''. You can have any number of preposi5onal phrases, for
example: ``the woman in a room on the chair in a room in the room sees the man''. We could have a
preposi5onal phrase modifying a verb as in ``the woman sleeps in the room'', but do not for this
coursework.
As an adjec5ve or preposi5onal phrase modifies a noun phrase, it can appear with the noun phrase in
either the subject or the object posi5on.
An adjec5ve or preposi5onal phrase is op5onal in the sense that not having them in a sentence results in
a sentence that is s5ll gramma5cal. However, there is a loss of meaning
As a hint about the grammar of adjec5ves and preposi5onal phrases in noun phrases, see the phrase
tree for sample sentences below. They indicate the gramma5cal structure of the categories and phrase
structure for adjec5ves and preposi5onal phrases in noun phrases; though somewhat complicated, it
shows the variety of structures. While the gramma5cal structure of jp and nbar are unfamiliar, we can
take them as given. Use these categories and phrase structures for your grammar. Given such input (and
similar), your parser should produce the same sort of output:
s(Tree, [the, woman, on, two, chairs, in, a, room, sees, two, tall, young, men], []).
Tree = s(np(det(the), nbar(n(woman)), pp(prep(on), np(det(two), nbar(n(chairs)), pp(prep(in),
np(det(a), nbar(n(room))))))), vp(v(sees), np(det(two), nbar(jp(adj(tall), jp(adj(young),
n(men)))))))
s(Tree, [the, woman, in, a, room, sees, two, young, men], []).
s(np(det(the),nbar(n(woman)),pp(prep(in),np(det(a),nbar(n(room))))),vp(v(sees),np(det(two),nb
ar(jp(adj(young),n(men))))))
All this said, there is a difference between ``ordinary'' preposi5onal phrases and a par5cular
preposi5onal phrase that appears in the passive. For our purposes, we will differen5ate them. See
below.
The Passive
Passive and ac5ve sentences are closely related:
the dog bites the woman. (ac5ve)
s(Tree,[the,dog, bites, the, woman],[]).
s(np(det(the),nbar(n(dog))),vp(v(bites),np(det(the),nbar(n(woman)))))
the woman is bicen by the dog. (passive)
s(Tree,[the,woman, is, bicen, by, the, dog],[]).
s(np(det(the),nbar(n(woman))),vp(aux(is),v(bicen),byPrepP(byPrep(by),np(det(the),nbar(n(dog))
))))
The passive is part of a much more widespread and diverse family of gramma5cal construc5ons called
diathesis alterna5ons. In diathesis alterna5ons, the arguments of the verb appear in alterna5ve posi5ons
yet with largely the same meaning. The passive and ac5ve sentences above mean the same thing, but
given in a different way and with some different rhetorical uses. Other examples of diathesis:
Direct-Indirect Object: A woman gives a book to a man; A woman gives a man a book
Causa5ve: The woman broke the chair; the chair was broken
We only consider the passive. While the past tense would be nicest, we have kept to the present tense
(doesn't really macer). There are several characteris5cs of the Passive construc5on in English:
• The posi5ons and case of the noun phrases change: what is the object (accusa5ve) NP in the
ac5ve sentences is the subject (nomina5ve) in the passive; what is the subject (nomina5ve and
'animate doer') NP in the ac5ve is in a par5cular preposi5onal phrase which represents the
'animate doer' of the ac5on.
she hires him.
s(Tree, [she,hires,him], []).
s(np(pro(she)),vp(v(hires),np(pro(him))))
he is hired by her.
s(Tree, [he,is,hired,by,her], []).
s(np(pro(he)),vp(aux(is),v(hired),byPrepP(byPrep(by),np(pro(her)))))
*her hires him.
s(Tree, [her,bites,him], []).
false.
*he is hired by she.
s(Tree, [he,is,hired,by,she], []).
false
• An ``auxiliary'' or ``helper'' verb (a form of ``to be’’ in this lexicon) is introduced in the passive.
Without the auxiliary, it might be read as a nominal, which is another macer.
*the woman bicen by the dog.
s(Tree,[the,woman, bicen, by, the, dog],[]).
false
• While the subject of the verb must be animate in the ac5ve, it need not be in the passive.
s(Tree, [the,woman,breaks,the,chair], []).
s(np(det(the),nbar(n(woman))),vp(v(breaks),np(det(the),nbar(n(chair)))))
s(Tree, [the,apple,breaks,the,chair], []).
false
s(Tree, [the,chair,is,broken,by,the,woman], []).
s(np(det(the),nbar(n(chair))),vp(aux(is),v(broken),byPrepP(byPrep(by),np(det(the),nbar(
n(woman))))))
• The verb in the ac5ve appears in a ``past par5ciple'' form in the passive.
The dog bites the woman.
The woman is bicen by the dog.
s(Tree, [the,woman,is,bites,by,the,dog], []).
false
• The par5cular preposi5onal phrase is op5onal without loss of meaning. This is in contrast with
dropping ``ordinary'' preposi5onal phrases.
The woman is bicen by the dog.
s(Tree,[the,woman,is,bicen],[]).
s(np(det(the),nbar(n(woman))),vp(aux(is),v(bicen)))
• Note the phrase structure. The auxiliary, the passive par5ciple, and the par5cular preposi5onal
phrase all appear together at the same ``level'' in the phrase structure in the VP. This is in
contrast to ordinary preposi5onal phrases which modify noun phrases.
The woman is bicen by the dog.
s(Tree, [the,woman,in,the,room,is,bicen,by,the,dog,in,the,room], []).
s(np(det(the),nbar(n(woman)),pp(prep(in),np(det(the),nbar(n(room))))),vp(aux(is),v(bic
en),byPrepP(byPrep(by),np(det(the),nbar(n(dog)),pp(prep(in),np(det(the),nbar(n(room))
))))))
The Lexicon
The lexicon should include all the following words that appear, where the components of each lexical
entry are as given. This is not the form that your code requires, but is a helpful hint.
The grammar must treat the features in the lexicon.
%%%%%%%%%%%% Lexicon %%%%%%%%%%%%%%
% The lexicon should include all the following words that appear, where the components of each lexical
% entry are as given. For clarity, the lexicon is given in the form that your code would require.
% The grammar must treat the features in the lexicon.
% Note that there are some lexical forms that your grammar would require, but are missing in the lis5ng
below.
%%% Pronouns %%%
% For pronouns, the informa5on appears in the following order: word, gramma5cal category (pronoun),
% number (singular/plural), gramma5cal person (1st, 2nd, or 3rd), and gramma5cal role (subject or
object)
lex(i,pro,singular,1,nom,ani).
lex(you,pro,singular,2,nom,ani).
lex(he,pro,singular,3,nom,ani).
lex(she,pro,singular,3,nom,ani).
lex(it,pro,singular,3,nom,ani).
lex(we,pro,plural,1,nom,ani).
lex(you,pro,plural,2,nom,ani).
lex(they,pro,plural,3,nom,ani).
lex(me,pro,singular,1,acc,ani).
lex(you,pro,singular,2,acc,ani).
lex(him,pro,singular,3,acc,ani).
lex(her,pro,singular,3,acc,ani).
lex(it,pro,singular,3,acc,ani).
lex(us,pro,plural,1,acc,ani).
lex(you,pro,plural,2,acc,ani).
lex(them,pro,plural,3,acc,ani).
%%% Common Nouns %%%
% For common nouns, the informa5on appears in the following order: word, gramma5cal category
(noun), number
lex(man,n,singular,_,_,ani).
lex(woman,n,singular,_,_,ani).
lex(dog,n,singular,_,_,ani).
lex(apple,n,singular,_,_,nani).
lex(chair,n,singular,_,_,nani).
lex(room,n,singular,_,_,nani).
% Thema5c rules we won't use.
%lex(X,n,singular,_,_,agent) :- lex(X,n,singular,_,_,ani).
%lex(X,n,singular,_,_,experiencer) :- lex(X,n,singular,_,_,ani).
lex(men,n,plural,_,_,ani).
lex(women,n,plural,_,_,ani).
lex(dogs,n,singular,_,_,ani).
lex(apples,n,plural,_,_,nani).
lex(chairs,n,plural,_,_,nani).
lex(rooms,n,plural,_,_,nani).
%%% Verbs %%%
% For verbs, the informa5on appears in the following order: word, gramma5cal category (verb), number
(singular/plural),
% gramma5cal person (1st, 2nd, 3rd)
%%% Transi5ve Verbs %%%
% Note that we do not have in this example Lexicon past par5ciples for every transi5ve verb:
% know, see, hire
lex(know,tv,singular,1,ani).
lex(know,tv,singular,2,ani).
lex(knows,tv,singular,3,ani).
lex(know,tv,plural,_,ani).
lex(see,tv,singular,1,ani).
lex(see,tv,singular,2,ani).
lex(sees,tv,singular,3,ani).
lex(see,tv,plural,_,ani).
lex(hire,tv,singular,1,ani).
lex(hire,tv,singular,2,ani).
lex(hires,tv,singular,3,ani).
lex(hire,tv,plural,_,ani).
lex(break,tv,singular,1,ani).
lex(break,tv,singular,2,ani).
lex(breaks,tv,singular,3,ani).
lex(break,tv,plural,_,ani).
lex(bite,tv,singular,1,ani).
lex(bite,tv,singular,2,ani).
lex(bites,tv,singular,3,ani).
lex(bite,tv,plural,_,ani).
%%% Past Par5ciple %%%
% These verb forms are used in the passive.
% Only the tv verbs can be passivized. We don't provide them all in this lis5ng,
% but as part of the coursework, you should add the pastPart of other tv verbs to the lexicon
% so as to recognise some of the sentences above and others.
% Note that verbs in the passive to not require animate subjects, though this is required of
% the transi5ve forms.
%
% You will need to add to the lexicon past par5ciple forms for all those verbs that are transi5ve above.
lex(broken,pastPart,singular,1,_).
lex(broken,pastPart,singular,2,_).
lex(broken,pastPart,singular,3,_).
lex(broken,pastPart,plural,_,_).
lex(bicen,pastPart,singular, 1, _).
lex(bicen,pastPart,singular, 2, _).
lex(bicen,pastPart,singular, 3, _).
lex(bicen,pastPart,plural, _, _).
%%% Intransi5ve verbs %%%
% These cannot go into the passive.
lex(fall,iv,singular,1,_).
lex(fall,iv,singular,2,_).
lex(falls,iv,singular,3,_).
lex(fall,iv,plural,_,_).
lex(sleep,iv,singular,1,ani).
lex(sleep,iv,singular,2,ani).
lex(sleeps,iv,singular,3,ani).
lex(sleep,iv,plural,_,ani).
%%% Auxiliary verbs (aux) for the passive %%%
% In this version of the lexicon, animacy of auxilary verbs is not necessary,
% though if the grammar is done in a different way, it might be.
lex(am,aux,singular,1).
lex(are,aux,singular,2).
lex(is,aux,singular,3).
lex(are,aux,plural,_).
%%% Determiners %%%
% For determiners, the informa5on appears in the following order: word, gramma5cal category
% (determiner), number
lex(the,det,_).
lex(a,det,singular).
lex(two,det,plural).
%%% Preposi5ons %%%
% For preposi5ons, the informa5on appears in the following order: word, gramma5cal category
(preposi5on)
lex(on,prep).
lex(in,prep).
lex(under,prep).
% We have a desigated category for the preposi5on for the passive by-phrase.
% When we connect to the grammar and used to recognised sentences, it will require
% an animate np. Consult the examples and discussion above.
lex(by,byPrep).
%%% Adjec5ves %%%
% For adjec5ves, the informa5on appears in the following order: word, gramma5cal category (adjec5ve)
lex(old,adj).
lex(young,adj).
lex(red,adj).
lex(short,adj).
lex(tall,adj).

请加QQ：99515681  邮箱：99515681@qq.com   WX：codinghelp

标签：