1. The dog barks loudly every mornings.
Phonological: Correct. Sounds fine when spoken.
Morphological: Incorrect. "Mornings" should be singular ("morning") to match the structure of the sentence.
Syntactic: Mostly correct, aside from the plural mismatch with "mornings."
Semantic: Correct. A dog can bark loudly in the morning.
Pragmatic: Natural if the morphological error is fixed.
2. A chair is running in the park.
Phonological: Correct. Sounds fine when spoken.
Morphological: Correct. Words are properly formed.
Syntactic: Correct. No grammatical issues.
Semantic: Incorrect. A chair cannot logically run.
Pragmatic: Unnatural unless used metaphorically or in a fictional context.
3. The baby cry for food.
Phonological: Correct. Sounds fine.
Morphological: Incorrect. "Cry" should be "cries" to agree with the singular subject "baby."
Syntactic: Incorrect due to the subject-verb agreement issue.
Semantic: Correct. Babies cry for food.
Pragmatic: Natural if the morphological error is fixed.
4. She can speaks three languages.
Phonological: Correct. Sounds fine.
Morphological: Incorrect. "Speaks" should be "speak" because the modal verb "can" requires the base form of the verb.
Syntactic: Incorrect due to the mismatch between "can" and "speaks."
Semantic: Correct. It’s logical for someone to speak multiple languages.
Pragmatic: Natural if the morphological error is fixed.
5. They went to the beach tomorrow.
Phonological: Correct. Sounds fine.
Morphological: Correct. Words are properly formed.
Syntactic: Incorrect. Mixing past tense ("went") with the future time marker ("tomorrow") violates tense agreement.
Semantic: Incorrect. It's illogical to go to the beach in the past on a future date.
Pragmatic: Confusing unless used intentionally for humor or emphasis.
6. Sometimes birds sings beautifully.
Phonological: Correct. Sounds fine.
Morphological: Incorrect. "Sings" should be "sing" to agree with the plural subject "birds."
Syntactic: Incorrect due to the subject-verb agreement issue.
Semantic: Correct. Birds can sing beautifully.
Pragmatic: Natural if the morphological error is fixed.
7. The car was so fast that it went slower than a bicycle.
Phonological: Correct. Sounds fine.
Morphological: Correct. Words are properly formed.
Syntactic: Correct. No grammatical issues.
Semantic: Incorrect. The sentence is logically contradictory (a fast car moving slower than a bicycle).
Pragmatic: Might work in a sarcastic or ironic context.
8. The sky is blue because it’s raining outside.
Phonological: Correct. Sounds fine.
Morphological: Correct. Words are properly formed.
Syntactic: Correct. No grammatical issues.
Semantic: Incorrect. The sky is usually gray or cloudy when it rains.
Pragmatic: Unnatural unless used metaphorically or in a poetic context.
9. A fish swims on the tree.
Phonological: Correct. Sounds fine.
Morphological: Correct. Words are properly formed.
Syntactic: Correct. No grammatical issues.
Semantic: Incorrect. Fish cannot swim on a tree; it violates real-world knowledge.
Pragmatic: Unnatural unless used in a metaphorical or surreal context.
10. John, who love books, is reading quietly.
Phonological: Correct. Sounds fine.
Morphological: Incorrect. "Love" should be "loves" to agree with
the singular subject "John."
Syntactic: Incorrect due to the subject-verb agreement issue.
Semantic: Correct. John can love books and read quietly.
Pragmatic: Natural if the morphological error is fixed.
In an Information Retrieval (IR) system, whether you prioritize precision or recall depends on the specific use case and the
consequences of missing or including irrelevant results. Let’s explore this:
1. Precision: When it Should Be High
Precision is the ratio of relevant documents retrieved to the total documents retrieved:
Precision=Relevant Retrieved DocumentsTotal Retrieved Documents\text{Precision} = \frac{\text{Relevant Retrieved Documents}}{\text{Total
Retrieved Documents}}Precision=Total Retrieved DocumentsRelevant Retrieved Documents
You want high precision when retrieving irrelevant documents is costly, harmful, or confusing.
Examples:
1. Legal Document Retrieval:
○ A lawyer searching for case precedents cannot afford to sift through many irrelevant cases.
○ High precision ensures that most retrieved cases are relevant to the query.
2. Medical Information Retrieval:
○ A doctor researching treatment options for a specific condition needs only highly relevant papers or studies.
○ Irrelevant documents might waste time or lead to misinformed decisions.
3. E-commerce Search Results:
○ A user searching for "wireless headphones" expects to see only relevant products.
○ Irrelevant results (e.g., "wired headphones" or unrelated electronics) might frustrate the user and lead to a poor
experience.
4. Question Answering Systems:
○ When users ask precise questions (e.g., "What year did the Apollo 11 mission land on the moon?"), irrelevant
results reduce trust in the system.
In all these cases, precision is critical because delivering irrelevant results diminishes the system's effectiveness or harms the
user's goals.
2. Recall: When it Should Be High
Recall is the ratio of relevant documents retrieved to the total relevant documents in the collection:
Recall=Relevant Retrieved DocumentsTotal Relevant Documents\text{Recall} = \frac{\text{Relevant Retrieved Documents}}{\text{Total
Relevant Documents}}Recall=Total Relevant DocumentsRelevant Retrieved Documents
You want high recall when missing relevant documents is unacceptable or has serious consequences.
Examples:
1. Criminal Investigations:
○ In retrieving evidence or surveillance data, investigators need to ensure that no relevant evidence is overlooked.
○ High recall ensures that as many relevant documents as possible are retrieved, even if some irrelevant ones are
included.
2. Medical Diagnosis Systems:
○ If a system retrieves medical studies or tests related to a rare disease, it’s critical to retrieve all potentially
relevant documents to avoid missing important information.
○ False negatives (missing relevant studies) can lead to incorrect diagnoses.
3. Academic Research:
○ A researcher performing a literature review wants to ensure they retrieve all relevant papers on a specific topic.
○ Missing important papers could lead to incomplete or biased conclusions.
4. Fraud Detection:
○ In identifying potentially fraudulent transactions, a system must flag all suspicious transactions, even at the cost
of false positives.
○ High recall ensures that no fraudulent activity is missed.
5. Search Engines for Specialized Domains:
○ For large-scale information systems like Google Scholar or patent search engines, high recall ensures users
can access all relevant information, even if it means sifting through irrelevant results.
Balancing Precision and Recall
In practice, precision and recall are often at odds:
● Increasing precision: Reduces irrelevant results but risks missing some relevant ones.
● Increasing recall: Retrieves more relevant results but includes more irrelevant ones.
To find a balance, you can use:
● F1-score: The harmonic mean of precision and recall, useful when both are equally important.
F1=2⋅Precision⋅RecallPrecision+RecallF1 = 2 \cdot \frac{\text{Precision} \cdot \text{Recall}}{\text{Precision} +
\text{Recall}}F1=2⋅Precision+RecallPrecision⋅Recall
● Domain-specific thresholds: Adjusting the importance of precision vs. recall based on the use case.