Natural language processing solutions 25-30% Flashcards

1
Q

Read OCR

A

Extract print and handwritten text including words, locations, and detected languages

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

Layout

A

Extract text and document layout elements like tables, selection marks, titles, section headings, and more

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

General

A

Extract key-value pairs in addition to text and document structure information

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

Health insurance card

A

Prebuilt: automate healthcare processes by extracting insurer , member, prescription, group number, and other key information from US health insurance cards

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

w-2

A

Prebuilt: process w2 forms to extract employee, employer, wage, and other information

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

invoice

A

Prebuilt: automate invoices

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

Receipt

A

Prebuilt: extract receipt data from receipts

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

Identity document

A

Prebuilt: extract identity fields from US driver licenses and international passports

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

Business card

A

Prebuilt: scan business cards to extract key fields and data into your applications

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

Form Recognizer prebuilt models

A

Health insurance car, W-2, invoice, receipt, identity document, business card

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

Form recognizer custom models

A

Custom extraction models, custom classification models, composed models

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

Translations services

A

Language detection, one to many translation, transliteration

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

Preconfigured features of a language understanding app

A

Summarization, named entity recognition, PII detection, key phrase extraction, sentiment analysis, language detection

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

Learned features of a language understanding app

A

conversational language understanding, custom-named entity recognition, custom text classification, question answering,

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

What three parameters will all conversational language understanding requests have

A

kind (type), parameters, analysis input

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

How to fix substitution errors?

A

add custom product and people names to the training data

17
Q

What type of error do you get with overlapping speakers?

A

deletion errors

18
Q

What type of error do people talking in the background give you?

A

insertion errors

19
Q

What is text independent verification?

A

The app will identify voices of people without a special verification phrase

20
Q

What is text dependent verification?

A

The app will identify voices with a passphrase

21
Q

What are the speech to text models?

A

base, custom, adapted

22
Q

When should you use a custom acoustic speech to text model?

A

Noisy areas

23
Q

When should you use a custom language speech to text model?

A

topics like biology, physics, radiology, product names, and custom acronyms

24
Q

What is the BLEU score?

A

An algorithm for evaluating the precision or accuracy of text that has been machine translated from one language to another.

25
What are the bounds of the BLEU score?
zero to one hundred
26
What range of scores indicates a high quality BLEU score?
40-60
27
What are not allowed in synonyms?
special characters like #