C H A P T E R 1 8
Intelligent Assistant
18-2
About the Assistant
templates may be supplied by the system or by your application. Some system-
supplied templates make use of lexical dictionaries, which are also supplied by the
system. For more information about lexical dictionaries, see Chapter 9,
"Recognition."
supplied templates make use of lexical dictionaries, which are also supplied by the
system. For more information about lexical dictionaries, see Chapter 9,
"Recognition."
Depending on the amount of information that parsing the input string provides, the
Assistant either attempts to complete a task or prompts the user to supply
additional information.
Assistant either attempts to complete a task or prompts the user to supply
additional information.
Input Strings
18
The Assistant is preprogrammed with a list of words representing tasks that the
built-in applications can perform. In addition to using these words, you can
program the Assistant to associate your own words with tasks your application
performs. The user cannot add words to the Assistant's vocabulary.
built-in applications can perform. In addition to using these words, you can
program the Assistant to associate your own words with tasks your application
performs. The user cannot add words to the Assistant's vocabulary.
You can associate multiple verbs with a single action; for example, the system
supplies a template that maps the words
supplies a template that maps the words
"call"
,
"dial"
,
"phone"
, and
"ring"
to the same task.
The Assistant uses some of the same dictionaries as the recognition system when
attempting to classify items in the input string. For example, it uses the system's
built-in lexical dictionaries to classify strings as phone numbers.
attempting to classify items in the input string. For example, it uses the system's
built-in lexical dictionaries to classify strings as phone numbers.
The word order in the input phrase is not significant-- for example, the input
phrase
phrase
"Bob fax"
produces the same result as the phrase
"Fax Bob"
. This
syntax-independent architecture allows easier localization of applications for
international audiences.
international audiences.
Note
The input string passed to the Assistant must not contain more
than 15 words. The Assistant does not attempt to interpret strings
containing more than 15 words.
than 15 words. The Assistant does not attempt to interpret strings
containing more than 15 words.
No Verb in Input String
18
If the Assistant cannot determine the user's intended action, it displays an Assist
slip that prompts the user to tap the Please picker for more options. The Please
picker allows the user to specify an action when the Assistant cannot determine one
by parsing the input string.
slip that prompts the user to tap the Please picker for more options. The Please
picker allows the user to specify an action when the Assistant cannot determine one
by parsing the input string.
For example, using the string
"Bob"
as partial input, the Assistant can perform a
number of actions: it can find Bob, fax Bob, call Bob, schedule a meeting with
Bob, and so on. However, this input string does not indicate which of these actions
is the user's actual intent. Figure 18-1 shows the Assist slip as it would appear if
the string
Bob, and so on. However, this input string does not indicate which of these actions
is the user's actual intent. Figure 18-1 shows the Assist slip as it would appear if
the string
"Bob"
were the only input provided to the Assistant.