tblLVST_NBest_Answers
Table Description:
Stores the NBest results for ASR Answers. Pretty typically, there will be only one NBest per interaction,
but for systems using NBest (n > 1), each interaction will have multple entries in this tables.
Table structure:
| Column Name |
Data Type |
Data Length |
Default Value |
See Also |
| NBestID |
INTEGER |
N/A |
N/A |
N/A |
| AnswerID |
INTEGER |
N/A |
N/A |
tblLVST_Answers |
| Rank |
INTEGER |
N/A |
0 |
N/A |
| AcousticModelScore |
FLOAT |
N/A |
-1 |
N/A |
| AverageWordScore |
FLOAT |
N/A |
-1 |
N/A |
| Noise |
FLOAT |
N/A |
-1 |
N/A |
| UtteranceScore |
FLOAT |
N/A |
-1 |
N/A |
Primary Keys:
NBestID
| Foreign Key |
Reference Table |
Reference Key |
| AnswerID |
tblLVST_Answers |
AnswerID |
Column Descriptions and Constraints:
| Column |
Description |
Constraint Name |
Constraint Description |
| NBestID |
Autoincrementing primary key. |
N/A |
N/A |
| AnswerID |
Foreign key to tblLVST_Answers. Identifies the answer this NBest result is part of.
|
N/A |
N/A |
| Rank |
Identifies the ordinal rank of this n-best result; a lower number indicates a higher rank,
i.e. 0 here indicates "First best", 1 indicates "second best", etc.
|
N/A |
N/A |
| AcousticModelScore |
The search probability of this NBest result. Care should be taken when comparing these numbers.
Nearly everything affects this, including the size of the grammar, the length of the utterance,
and the quality of acoustic models and dictionary. This number is typically *NOT* directly comparable
to any other result, and the normalization required to make them comparable can be quite complex.
Make sure you understand all of the factors before drawing conclusions between scores. If the field
is set to -1, no acoustic model score was available.
|
N/A |
N/A |
| AverageWordScore |
The average confidence score for all words in the raw text result. This number is very sensitive
to outliers, like any average. Be careful about affording it too much importance. If this field
is set to -1, the average word score was not avialable.
|
N/A |
N/A |
| Noise |
Roughly, a signal-to-noise ratio (SNR). Measures how much of the result was attributed to noise
versus actual speech. It has no noise score if it's value is -1.
|
N/A |
N/A |
| UtteranceScore |
This combined measure is a look at how good a result the entire *utterance* is, based on combining
scores from acoustics, confidence measures, grammars, etc. If the utterance score is not avialable,
this field is set to -1.
|
N/A |
N/A |
Table Constraints:
| Constraint Name |
Constraint Description |
Columns Related |
| N/A |
N/A |
N/A |