5. (6 marks) Information Extraction and Summarisation (a) (3 marks) Named entity

Question

5. (6 marks) Information Extraction and Summarisation (a) (3 marks) Named entity recognition is a typical task in information extraction. The fol- lowing named entity types were defined in the Message Understanding Conference (MUC) Identify three types that can be detected with regular expressions reasonably well, and three types that cannot be detected with regular expressions satisfactorily. Justify your 1. Dates. 3. Duration. 4. Money 5. Measure. 7. Cardinal number 9. Organisation name. 10. Location name.

Explanation / Answer

5 a)

Three types that can be identified using regular expressions:

1.Dates : Dates can easily be identified using regular expressions as they have defined set of numbers and a limited set of seperators(/ or -)

2.Times : Times has a fixed format which can be formed using regular expression. Eg : 12:00:00 This can easily be identified using a regular expression.

3. Percentage : Percentage also has a fixed format such as 95% or 91.2%. A regular expression like ^[0-9]*[.][0-9]*% will identify a percentage.

Three types that cannot be identified using regular expressions:

Personal name, organisational name,location name - It is impossible to identify any proper noun using regular expression as the word could evaluate to all possiblities which might include those that does not fall under this category.

5 b.

1. Frequency-keyword : Score the words based on its frequency.

2. Biased - keyword : Giving more weightage to certain keywords over others.

3. Location : Scoring a word based on its location in a sentence. Eg: The word appearing at subject location get more score than the same word appearing as an object.

Related Questions

Navigate

• Browse (All)
• Browse 5
• Subjects

---

---

Integrity-first tutoring: explanations and feedback only — we do not complete graded work. Learn more.