NEW
Font size
S
M
L
XL
WorksheetsWeb Mining 6.3
Total questions: 67
Worksheet time: 34mins
Name
Class
Date
1.
In the context of forums, why is entity detection and assignment needed?
a)
To identify which entity the comments are aimed at
b)
To compress text to reduce storage cost
c)
To translate posts into another language
d)
To remove all neutral sentences from a thread
e)
To convert images into text
2.
Under the sentiment consistency example, what relationship is inferred from the comparison statement?
a)
Camera-A is better than Camera-B
b)
Camera-B is better than Camera-A
c)
Camera-A and Camera-B are equal in quality
d)
Neither camera is being evaluated
e)
The sentence is unrelated to any camera
3.
In the problem statement, what does the entity set epsilon represent for a sentence?
a)
The set of entities that the sentence reviews
b)
The set of all posts in the thread
c)
The list of all words in the sentence
d)
The set of all possible sentiment labels
e)
The set of all users in the forum
4.
According to the slides, sentences involving more than one entity are usually what type of sentences?
a)
Comparative sentences
b)
Greeting sentences
c)
Spam detection sentences
d)
Topic modeling sentences
e)
Language identification sentences
5.
In the thread-level formulation, what is Task 2 in the problem statement?
a)
Assign each sentence with one or several entities in E
b)
Detect the polarity of every opinion word in a dictionary
c)
Cluster all posts into topics
d)
Remove stopwords from each post
e)
Predict the author of each sentence
6.
In the unsupervised entity detection method described, what is the second main step?
a)
Sequential pattern mining
b)
Prepare data
c)
Extract candidate
d)
Filter candidates
e)
Train a supervised classifier
7.
During data preparation, what is done to entity names found in sentences?
a)
They are replaced with a generic name such as ENTITYXYZ
b)
They are deleted from the sentence
c)
They are translated into Vietnamese
d)
They are converted into numbers only
e)
They are split into characters and reversed
8.
In sequential pattern mining, which constraint is explicitly required for the mined patterns?
a)
Patterns must contain POS and ENTITYXYZ
b)
Patterns must contain at least three different entities
c)
Patterns must contain only adjectives
d)
Patterns must contain no part-of-speech tags
e)
Patterns must be exactly length 10
9.
What is the purpose of the candidate extraction step in entity detection?
a)
Search for entities that match the generated patterns
b)
Assign sentiment labels to clauses
c)
Merge duplicate threads across domains
d)
Detect spam users in the forum
e)
Summarize each post into one sentence
10.
In filtering candidates, why would a token like accessories/CD be excluded?
a)
Because its POS tag differs from the POS most popular for that candidate
b)
Because it contains too many characters
c)
Because it appears only in product reviews
d)
Because it is always a positive opinion word
e)
Because it is a stopword
11.
Which of the following is listed as a syntax pattern to find competing brands?
a)
A vs B
b)
A inside B
c)
A before B
d)
A because of B
e)
A equals not B
12.
Which type of comparison sentence is illustrated by 'Camera-X's battery life is the longest'?
a)
Superlative
b)
Equal
c)
Non-comparable
d)
Comparative
e)
Greeting
13.
If an entity e first appears in sentence s0 and the next sentence s1 is a normal sentence, how is s1 handled in sentiment consistency?
a)
s1 is assigned to entity e
b)
s1 is ignored as unrelated
c)
s1 is assigned to a new entity introduced later
d)
s1 is always assigned to the worse entity
e)
s1 is always assigned to all entities in the thread
14.
If s0 is a superlative sentence and s1 is a normal sentence, which entity is s1 assigned to?
a)
The best entity mentioned in s0
b)
The worst entity mentioned in s0
c)
The first entity mentioned anywhere in the thread
d)
No entity because superlatives block assignment
e)
A random entity to balance the dataset
15.
In the algorithm slide, what is opinion() used for?
a)
Determine emotions in normal sentences
b)
Detect entities using sequential patterns
c)
Generate brand-model pairs
d)
Compute precision and recall in evaluation
e)
Find synonyms and antonyms in WordNet
16.
In sentiment analysis, which cue is explicitly mentioned as handling contrast between clauses?
a)
The clause connector 'but'
b)
The presence of digits in a sentence
c)
The use of passive voice
d)
The number of commas in a post
e)
The presence of emojis only
17.
In the specification language, what does the example 'like[VB] => Po' illustrate?
a)
A rule that maps a word with a POS tag to an outcome label
b)
A rule that removes all verbs from a sentence
c)
A rule that converts adjectives into nouns
d)
A rule that replaces product names with ENTITYXYZ
e)
A rule that detects comparative sentences only
18.
In the example slide, what rule turns 'too slow' from neutral to negative evaluation?
a)
too + neutral adjective at target makes the outcome negative
b)
but + adjective always makes the outcome positive
c)
not + adjective always makes the outcome neutral
d)
too + adjective always makes the outcome positive
e)
slow + verb always makes the outcome neutral
19.
Which pattern form is listed for analyzing comparative sentences?
a)
product name + comparison word + product name
b)
product name + greeting + user name
c)
topic + hashtag + URL
d)
entity set + epsilon + thread id
e)
brand model + serial number + price
20.
According to the rules for comparative expressions, what is the polarity of 'less + positive'?
a)
Negative
b)
Positive
c)
Neutral
d)
Always depends on the product name
e)
Undefined because less cannot modify positive words
21.
Which dataset is described as covering Plasma/LCD TVs, Projectors and DVD players?
a)
AVSforums
b)
HowardForums
c)
Epinions
d)
WordNet
e)
Wikipedia
22.
What does Baseline1 do when the current sentence does not contain an entity?
a)
Get the last entity of the previous sentence
b)
Get the most frequent entity in the whole thread
c)
Get the first entity in the current sentence
d)
Discard the sentence as irrelevant
e)
Introduce a new entity placeholder
23.
In the result evaluation table for entity discovery, which metrics are reported?
a)
Recall and precision (and an F measure)
b)
Accuracy and latency only
c)
BLEU and ROUGE scores
d)
Perplexity and cross-entropy
e)
Number of pixels and screen resolution
24.
According to the slides, what is the more common form of user product reviews on the Internet?
a)
Direct reviews
b)
Comparative sentences
c)
Only questions without opinions
d)
Only links to external sites
e)
Only anonymous ratings with no text
25.
What challenge is highlighted about comparative sentences in the problem statement?
a)
Many do not have a direct comparison word and the same word can change emotion by context
b)
They always contain three or more products
c)
They never mention product features
d)
They are always written in a single clause
e)
They can only be analyzed with images
26.
What external resource is mentioned for identifying opinion orientation from context?
a)
Epinions.com
b)
HowardForums
c)
AVSforums
d)
WordNet
e)
Google Scholar
27.
In the relation representation <C, F, e1, e2, type>, what does F denote?
a)
Feature
b)
Forum
c)
Frequency
d)
File format
e)
False positive rate
28.
In the Pros and Cons lists, what does an entry like 'great photos <photo>' represent?
a)
A positive comment about a feature category
b)
A negative comparison between two entities
c)
A greeting sentence in a thread
d)
A stopword pattern for filtering candidates
e)
A rule for detecting POS tags
29.
For Type 1 (-er/-est) comparisons, if the comparison word C expresses positive emotion, which entity is chosen as better?
a)
e1
b)
e2
c)
Neither entity
d)
Both entities equally
e)
A new entity e3
30.
When both the comparison word C and feature F have no sentiment, what decision rule is used?
a)
Choose e1 if OSAP(F,C) is greater than OSAN(F,C); otherwise choose e2
b)
Always choose e1 because C is comparative
c)
Always choose e2 to avoid positive bias
d)
Randomly choose e1 or e2
e)
Choose the entity that appears last in the thread
31.
What lexical resource is used to obtain synonyms and antonyms when computing association counts?
a)
WordNet
b)
Epinions.com
c)
HowardForums
d)
AVSforums
e)
PageRank
32.
If the comparison word C exhibits a feature (for example, smaller), how is its preference inferred?
a)
Count how often C occurs in Pros and in Cons, then choose the larger value
b)
Treat all such comparisons as neutral
c)
Always prefer the larger entity
d)
Ignore Pros and Cons and use only POS tags
e)
Use only the first sentence in the post as context
33.
Which construction is explicitly listed as a negative case in Type 2 comparisons?
a)
A negated comparison such as 'not longer than'
b)
A greeting such as 'hello everyone'
c)
A question such as 'what is the price'
d)
A URL-only sentence
e)
A sentence with no adjectives
34.
The evaluation data is described as including comparisons across which kinds of items?
a)
A mix including cameras, DVD players, and brand or company pairs like Intel vs AMD
b)
Only movie reviews
c)
Only plasma TV reviews
d)
Only posts about one specific camera model
e)
Only grammar examples with POS tags
35.
In the evaluation slide, what does the simple baseline do to assign entities?
a)
Always takes the first entity
b)
Always takes the last entity
c)
Always assigns no entity
d)
Always assigns a random entity
e)
Always assigns the most recent verb
36.
If you already know the set of entities in a domain and you need to label each sentence with the relevant entity or entities, which task is this?
a)
Entity assignment
b)
Entity detection
c)
Sequential pattern mining
d)
Opinion word extraction
e)
Candidate filtering
37.
Why does replacing each entity mention with a generic token like ENTITYXYZ help in the unsupervised entity detection method?
a)
It lets patterns capture shared context around entities instead of memorizing specific names
b)
It makes every sentence contain exactly one entity
c)
It guarantees that all extracted candidates are brands only
d)
It removes the need for POS tagging
e)
It converts comparative sentences into normal sentences
38.
A mined sequence contains ENTITYXYZ but does not include any POS information. Based on the constraints, what should happen to this sequence?
a)
It should be rejected because patterns must contain POS and ENTITYXYZ
b)
It should be accepted because ENTITYXYZ is sufficient
c)
It should be accepted only if the sequence length is 1
d)
It should be used only for superlative sentences
e)
It should be treated as a negative pattern automatically
39.
A candidate entity appears in text with two POS tags, but one POS tag is much more common for that candidate. How does the filtering step handle mismatched POS cases?
a)
Exclude occurrences whose POS differs from the most popular POS for that candidate
b)
Keep all occurrences to maximize recall
c)
Convert all POS tags to NN to standardize
d)
Exclude the candidate entirely regardless of frequency
e)
Assign the candidate to both POS tags equally
40.
Under sentiment consistency, if an entity first appears in s0 and s1 is a comparison sentence, what is implied about s1?
a)
The entity from s0 is compared with a new entity that needs to be introduced
b)
s1 must be ignored because comparisons cannot be assigned
c)
s1 should be assigned to the worse entity by default
d)
s1 always evaluates both entities equally
e)
s1 is treated as unrelated greeting text
41.
Suppose s0 is a comparative sentence and s1 expresses sentiment but mentions no entity. How is s1 assigned according to the sentiment consistency rule?
a)
Assign s1 to the better or worse entity from s0 depending on whether s1 is positive or negative
b)
Assign s1 to a new entity because no entity is mentioned
c)
Assign s1 to every entity in the thread
d)
Do not assign s1 because it is missing an entity
e)
Assign s1 to the last noun in s1
42.
If s0 is an equal or non-comparable sentence and s1 is a normal sentence with no clear entity, what assignment choice is made?
a)
Assign s1 to the entity that came before s0
b)
Assign s1 to both entities in s0
c)
Assign s1 to the better entity in s0
d)
Assign s1 to a new entity placeholder
e)
Assign s1 to the most frequent entity in the domain
43.
In the algorithm, if a sentence has no explicit entity and the previous sentence is not comparative, which entity is used?
a)
The entity assigned to the previous sentence
b)
The entity with the highest precision in evaluation
c)
The first entity in the thread
d)
No entity because assignment requires explicit mention
e)
A random entity to avoid bias
44.
A sentence says 'The picture quality is great, but not the battery life.' What does the cue 'but' suggest for sentiment handling?
a)
Different clauses may carry different sentiment targets, so sentiment may flip across the contrast
b)
The whole sentence is always positive
c)
The whole sentence is always negative
d)
The sentence should be discarded as neutral
e)
The sentence must be treated as a comparative sentence
45.
In the specification language, which label is used to mark the contrast connector in clause handling?
a)
But
b)
Po
c)
Ne
d)
Neu
e)
Ng
46.
Which pattern best matches a comparative sentence where a pronoun is compared to a product name using a comparison word?
a)
pronoun + compkey + product name
b)
product name + superkey
c)
as + adjective + as
d)
product name + greeting + product name
e)
product name + negation + stopword
47.
Given the rule set for more/most and less/least, what is the overall polarity of 'less reliable' if 'reliable' is treated as positive?
a)
Negative
b)
Positive
c)
Neutral
d)
Always depends on the entity order
e)
Always depends on whether the sentence has 'but'
48.
For Type 1 (-er/-est) cases where the feature is negative, what happens in a sentence like 'Car X generates more noise than Car Y'?
a)
Car Y is chosen as better because the comparative increases a negative feature for Car X
b)
Car X is chosen as better because it is mentioned first
c)
Both are chosen as better because noise is subjective
d)
No entity can be chosen because there is no comparison word
e)
Car X is chosen as better because more always means positive
49.
When both the comparison word C and feature F have no sentiment, how is the better entity decided?
a)
Compare OSAP(F,C) and OSAN(F,C) and choose e1 only if OSAP is larger
b)
Choose e1 whenever the sentence contains any adjective
c)
Choose e2 whenever the sentence contains any adjective
d)
Choose the entity with the longer name
e)
Choose the entity that appears in the Pros list less often
50.
If the comparison word describes a feature like 'smaller', and that word occurs more often in Cons than in Pros, what does the method infer?
a)
The feature is likely undesirable in the domain
b)
The feature is always desirable regardless of counts
c)
The feature must be neutral because size is not a sentiment
d)
The feature only applies to superlatives
e)
The feature cannot be evaluated without a direct comparison word
51.
The slides note that taking whole sentences as context can bring in irrelevant information. What context is proposed instead?
a)
Evaluated entity plus the comparison word
b)
All sentences in the thread
c)
Only greeting sentences
d)
Only the last sentence in each post
e)
Only the title of the forum thread
52.
How does using Epinions.com help infer opinion orientation from context?
a)
By checking whether a context appears more often in positive or negative comments
b)
By providing the correct entity names for every post
c)
By supplying POS tags for every word
d)
By converting comparison sentences into direct reviews
e)
By removing all negation words automatically
53.
If the previous sentence contains two entities 'A ... B' and the current sentence has no entity, which baseline would most likely pick B?
a)
Baseline1
b)
Baseline2
c)
CRF
d)
NET
e)
ED (k-com)
54.
What is the main difference between ED (k-com) and ED (unk-com) in the evaluation setup?
a)
ED (k-com) is given comparative sentences, while ED (unk-com) must detect them
b)
ED (k-com) uses no entities, while ED (unk-com) uses all entities
c)
ED (k-com) works only on AVSforums, while ED (unk-com) works only on HowardForums
d)
ED (k-com) uses WordNet, while ED (unk-com) uses no lexical resources
e)
ED (k-com) is supervised, while ED (unk-com) is unsupervised
55.
Consider three sentences in one post: (1) 'I bought Camera-A yesterday.' (2) 'Camera-A and Camera-B are of the same size.' (3) 'It is very good.' According to the sentiment consistency rules, which entity should sentence (3) be assigned to?
a)
Camera-A, because it came before the equal sentence
b)
Camera-B, because it is mentioned second in the equal sentence
c)
Both Camera-A and Camera-B, because equal sentences force dual assignment
d)
No entity, because equal sentences block later assignment
e)
A new entity, because 'it' is ambiguous
56.
A comparative sentence identifies a superior entity and an inferior entity. The next sentence has no entity mention. If that next sentence begins with a negative clause, which entity does the algorithm assign it to?
a)
The inferior entity from the previous comparative sentence
b)
The superior entity from the previous comparative sentence
c)
Both entities from the previous comparative sentence
d)
The first entity mentioned in the thread
e)
No entity because the sentence has no explicit entity
57.
The comparative pattern 'as + adjective + as' has exceptions such as 'as long as' and 'as far as'. What is the most plausible reason for excluding these exceptions from comparative detection?
a)
They often function as fixed phrases rather than true comparisons between entities
b)
They always contain product names, which breaks the pattern
c)
They only appear in negative reviews, so they add bias
d)
They require POS tags that are unavailable
e)
They can only be detected in images, not in text
58.
The slides note that the same comparison word can express different emotion depending on context (for example, 'longer' in battery life vs execution time). What determines whether the comparison is positive or negative in such cases?
a)
The feature being compared and its domain meaning
b)
The font style used in the review text
c)
The order of entities in the sentence only
d)
Whether the sentence includes a greeting
e)
Whether the review is posted on a forum or a product page only
59.
Suppose you use only the context 'evaluated entity + comparison word' instead of whole sentences. Which remaining problem still needs to be solved according to the slides?
a)
How to identify the emotion expressed by that context, potentially using external knowledge
b)
How to replace all entities with ENTITYXYZ
c)
How to compute POS tags without a tagger
d)
How to remove all comparative sentences from the dataset
e)
How to ensure every sentence contains exactly one entity
60.
Using the comparative polarity rules, what is the polarity of the phrase 'less ugly' if 'ugly' is treated as negative?
a)
Positive
b)
Negative
c)
Neutral
d)
Always undefined because ugly is subjective
e)
Always depends on whether 'but' appears later
61.
Apply the Type 1 (-er/-est) decision rules to: 'Car X has less noise than Car Y.' If 'noise' is treated as a negative feature, which entity should be chosen as better?
a)
Car X, because the comparison reduces a negative feature
b)
Car Y, because it is mentioned second
c)
Car X, because less always implies negative sentiment
d)
Car Y, because any mention of noise implies Car X is worse
e)
Neither, because there is no explicit opinion word like good or bad
62.
When both the comparison word C and feature F have no sentiment, the method compares OSAP(F,C) and OSAN(F,C). If OSAP(F,C) equals OSAN(F,C), what choice is made?
a)
Choose e2
b)
Choose e1
c)
Choose both e1 and e2
d)
Choose the entity that appears in Pros more often
e)
Choose the entity that appears first in the thread
63.
Which component is explicitly counted as part of OSAP(F,C) in the slides' description?
a)
How often C (and synonyms) and F (and synonyms) appear together in Pros
b)
How often C and F appear together anywhere in the thread regardless of polarity
c)
How often C appears with antonyms of F in Pros
d)
How often negation words appear near C in Pros
e)
How often product names appear near C in Pros
64.
In the entity detection pipeline, sequential patterns are mined from windows around ENTITYXYZ. What is a likely failure case if the relevant context for an entity lies outside the chosen window?
a)
The pattern mining may miss useful context and fail to extract that entity as a candidate
b)
The algorithm will always increase precision because less context is better
c)
The algorithm will always detect more competing brands
d)
The algorithm will always assign sentiment correctly due to sentiment consistency
e)
The filtering step will automatically recover the missing context
65.
The slides list 'not' and also 'not only ... but also' under negative cues. What does this imply about naive negation handling in sentiment analysis?
a)
Negation cues can be multi-word constructions, so looking for a single word 'not' may be insufficient
b)
Negation never affects sentiment orientation in text
c)
Negation always flips the sentiment of the whole sentence
d)
Negation can be ignored if the sentence contains a comparative word
e)
Negation is only relevant in superlative sentences
66.
In the specification language, outcomes include Po, Ne, Neu, Ng, and But. If a rule flags a token as But, how should a downstream sentiment analyzer most reasonably use it?
a)
Treat it as a signal of contrast that may change sentiment focus across clauses
b)
Treat it as a positive sentiment label for the current entity
c)
Treat it as a negative sentiment label for the current entity
d)
Drop the sentence because contrast makes sentiment undefined
e)
Assume the sentence is a superlative comparison
67.
Baseline1 uses the last entity of the previous sentence when the current sentence has no entity, while Baseline2 uses the first. In a post with many comparative sentences mentioning two entities, what systematic error can either baseline introduce?
a)
It can consistently attach follow-up sentiment to the wrong entity when entity order differs from the true target
b)
It can always outperform ED methods because it is simpler
c)
It can only fail on threads with more than five posts
d)
It can only fail when entities are explicitly mentioned
e)
It eliminates the need to detect comparative sentences
Reset
