Diversify utterances

[deejcunningham]

Currently only responds to the words "score" and a valid MLB team in the proper format: "D-backs" not "Diamondbacks", "Athletics" not "A's", "Cardinals" not "Cards", etc.

Acceptable utterances should be diversified to include all possible natural language variations for any given intent.

[deejcunningham]

Running through some utterances today...

First, diversifying natural language utterances: Possible natural language utterances for a LIVE score are:

• "What is the TEAM score"
• "What is the score of the TEAM game"
• "Who is winning|losing the TEAM game"
• "Are the TEAM winning|losing"
• "How are the TEAM doing" (may conflict with possible planned intent for reporting standings)

With the addition of context (a TEAM being implied based on previous interaction), this becomes possible:

• "What is the score"

Possible natural language utterances for a PAST score are:

• "What was the TEAM score [TIME] against [OPPONENT]"
• "What was the score of the TEAM game [TIME] against [OPPONENT]"
• "Who won|lost the TEAM game [TIME] against [OPPONENT]"
• "Did the TEAM win|lose [TIME] against [OPPONENT]"
• "How did the TEAM do [TIME] against [OPPONENT]"

[TIME] is an optional parameter that defaults to latest final score, but could be things like "earlier|today", "yesterday" or a day of the week with "last" implied beforehand "last|Wednesday", or even a raw date "June 15th". Will likely start with most commonly asked for times first "earlier|today", "yesterday", and day of week "last|Wednesday". Specific dates seem unlikely to be useful for the vast majority of users.

[OPPONENT] is another optional parameter that would default to the latest game against that opposing team or possibly the result of a series (but that is more in line with a standings intent to be implemented at a later date).

[deejcunningham]

(This comment was edited to remove some team names)

Team names should be updated and diversified. I believe the current STT engine is DeepSpeech which should be capable of handling all of these pronouns. Either way, I have tried to find possible 'nicknames' for MLB teams that a user may use instead of the full name. They should all eventually function equivalently:

Baltimore Orioles

• "Orioles"
• "O's"
• "Birds" XXX

Boston Red Sox

• "Red Sox"
• "Sox" XXX

New York Yankees

• "Yanks"
• "Bronx Bombers"
• "Bombers"
• "Evil Empire"

Tampa Bay Rays

• "Rays"
• "Devil Rays"

Toronto Blue Jays

• "Blue Jays"
• "Jays"

Chicago White Sox

• "White Sox"
• "Sox" XXX

Cleveland Indians

• "Indians"
• "Tribe"
• "Wahoos"

Detroit Tigers

• "Tigers"
• "Tiges"

Kansas City Royals

• "Royals"

Minnesota Twins

• "Twins"

Houston Astros

• "Astros"
• "'Stros"

Los Angeles Angels

• "Angels"
• "Halos"

Oakland Athletics

• "Athletics"
• "A's"

Seattle Mariners

• "Mariners"
• "M's"

Texas Rangers

• "Rangers"

Atlanta Braves

• "Braves"
• "Bravos"

Miami Marlins

• "Marlins"
• "Fish"

New York Mets

• "Mets"

Philadelphia Phillies

• "Phillies"
• "Phils"

Washington Nationals

• "Nationals"
• "Nats"

Chicago Cubs

• "Cubs"
• "Cubbies"

Cincinnati Reds

• "Reds"

Milwaukee Brewers

• "Brewers"

Pittsburgh Pirates

• "Pirates"
• "Bucs"
• "Buccos"

St. Louis Cardinals

• "Cardinals"
• "Cards"
• "Red Birds"
• "Birds" XXX

Arizona Diamondbacks

• "Diamondbacks"
• "D-backs"
• "Snakes"
• "Rattlesnakes"

Colorado Rockies

• "Rockies"
• "Rox"

Los Angeles Dodgers

• "Dodgers"

San Diego Padres

• "Padres"
• "Friars"

San Francisco Giants

• "Giants"

There are some conflicts. "Sox" for both the Red Sox and White Sox. "Birds" for the Orioles and Cardinals. These will not function until I build in some context so Mycroft can ask for a clarification (and hopefully remember that preference for the future)

Some thoughts on team names:

• It would be amazing to be able to attempt differentiation based on accent - "Sawx" vs "Socks". Maybe I should check out Padatious.
• Debating being able to refer to teams based on their city name. Some teams are not from a city: "Texas" Rangers even though they play in Arlington (which is really more Dallas-Fort Worth-Arlington anyways). ALSO, I would eventually like to integrate other sports (American football, etc) and many cities have multiple teams in different sports (though this could be solved by Mycroft asking the user to specify a sport/team), most people ask about their home team and wouldn't do so using the city name, not to mention some cities have more than one baseball team - e.g. New York, LA, Chicago.
• There is an opportunity here to have some fun for baseball nerds. E.g. the Dodgers have (at various times in their existence) been called the "Atlantics", "Bridegrooms", "Grooms", "Suberbas", "Robins", and "Trolley Dodgers". Will likely not implement this because it gets complicated VERY quickly - i.e. The Baltimore Orioles were founded in 1901 as the Milwaukee Brewers before moving to St. Louis and becoming the Browns after only one year, and then finally moving to Baltimore to become the Orioles in 1953. The Orioles name has also been used by several previous major and minor league clubs in Baltimore, including the team that moved to New York to become the Yankees.

[deejcunningham]

Doing some voice testing today and ran across the following:

Said "What's the O's score" - asking about the Orioles; Mycroft detects "what's the o's score", falls back on the Wolfram Alpha skill and reports "20 s o'". This also occurs with the "a's" and "m's".

Mycroft fails to detect "Tiges", "'Stros" and "Cubbies" - instead detecting "tigers", "astros" and "cubs". It reports the correct information anyways so not a big deal, but will keep an eye on these three.

Mycroft detects "redbirds" instead of "red birds", so that has been added.

Mycroft sometimes detects "dbacks" (in addition to normally detecting "d-backs") so "dbacks" has been added as well.

When speaking naturally, Mycroft sometimes drops the tailing s - i.e. in "Cards", "Snakes", "Wahoos". This is likely because the next word is usually "score" but maybe I should consider adding singular versions as valid names?

Mycroft detects "rock's" (and sometimes "rockets") instead of "Rox" when asking about the Rockies. Tried to add it as a valid reference, but Mycroft fails to make the skill connection. This (plus the problems with the 'letter' teams) leads me to believe that there is an issue when Mycroft detects the team name as containing an apostrophe (or apostrophe then the letter s). I will have to investigate this more.