Response: Finally, there are Responses. There is a set of possible Responses for each Keyphrase you have. When the Bot determines the most important Keyphrase, it will randomly select one of the corresponding Responses. This randomness is designed to prevent repetition. It is good to have several responses for each Keyphrase, for that increases the unpredictability of your Bot's Personality.

Plug-Ins: Plug-ins are ways of including specific or random information in your Keyphrases and Responses. There is a wide variety of plug-ins which will be covered as they come up. As an example, the Response "I like (animal)s" will appear as "I like eagles" or "I like horses".

These are the basics of your interface in The Personality Forge. Start with this "Beginner" section and come back to learn the others when you're comfortable with the basics.



Chapter 2: An Example Personality

Here is an example that shows some important Keyphrases along with some Responses. This is a very small selection from one of my own AI Personalities. These are just examples. You will probably discover some ways of reacting that I havent, and variety is the spice of any personality. In each category I have included only 2-3 responses, though in actuality there are many more than those shown. Some other important Keyphrases to use are: my, your, sorry, or, what, when, who, why, and where, which facilitate the ins and outs of conversation and memory usage. Please do not copy the following Responses- they are just meant as examples.

Language Center

NOTE: Many people have copied the above Responses into their own Bot and as a result, too many Bots are saying the same thing. In the spirit of originality and making things interesting, please do not copy the above Responses into your Bot. The Keyphrases are fine to have, but make the Responses unique.

Chapter 3: Building Your Bot- Keyphrases & Responses

Creating a New Bot
You can create a new Bot by going to "My Bots" and clicking "New Bot". You Bot will start out with very basic language skills- the same as other new Bots. A series of special "xKeyphrases" are created for you. These are central to any conversation and cannot be deleted. The responses associated with them can and should be changed to offer variety, personality, and unpredictability. XKeyphrases will be explained in more detail below. We'll start out with regular Keyphrases here.

You will do all your work on your Bot in the Language Center, which appears as a link under your Bot's name on the "My Bots" page.

Keyphrases
Creating New Keyphrases: At the bottom of your Language Center page is an area that can be used to add new Keyphrases and Responses. You can start by adding 7 Responses, but you can add more later.

Editing Keyphrases: To edit a Keyphrase, click the name of the Keyphrase in the Language Center.

Keyphrase Advice: Keyphrases are really where your Bot's intelligence will come out. A Bot with the following Keyphrase: "do you", "what do you", "do you like" and "do you think" will respond much more intelligently than a Bot that only has "do you". If you see your Bot missing a conversational cue, then look into their Keyphrase and add a Keyphrase that would be triggered by that cue. Start out with basic Keyphrases. Begin with "you", then add "you are", "you * me", "you like", "why do you", "did you" etc. Some of the most popular Keyphrases are: I, you, my, and your. Then you might go on to: how, what, when, where, why, and who. Then possibly: sorry, thanks, if, was, or, is, and so on. Then if your Bot brings up certain topics, such as skydiving, you might want to add Keyphrases relating to that.

• Contractions are not used in Keyphrases. Each contraction a Bot receives will be broken down into its basic parts. For example, if someone says "I'm here. I can't get any closer." Then the Bot will translate it to "I am here. I can not get any closer." So make sure that your Keyphrases do not contain contractions. Sometimes the result will seem strange, such as a Keyphrase of "why do not you" to recognize "why don't you" but this how to correctly phrase it.

• Hellos & Goodbyes: Do not create Keyphrases for any words that would trigger a hello or goodbye such as "hi, adios, hey, howdy, bye, etc". Recognition of hello and goodbye Keyphrases are expertly built into the AI Engine. Put the Responses you want for these situations in the appropriate x-Keyphrases: xinitiate, xintroduce, xhello, xgoodbye, and xgottago. (see below)

1) Word Placement Bonus: Keyphrases are automatically ranked according to their position in a sentence. If a Keyphrase is at the beginning of a sentence, it is ranked higher. If, for example, you have the Keyphrases "what" and "my" created, and a message comes in saying "What is my favorite drink?", then the "what" Keyphrase will be triggered because it is closer to the beginning of the sentence. The current value for this Ranking Bonus is 12. The first word in a sentence gets a +12 rank, the second gets a +6 rank, the third gets a +4 rank, the fourth gets a +3 rank, the fifth through the eigth gets a +2 rank, and anything beyond that gets a +1 rank.

2) Sentence Order Bonus: The final sentence is generally the best one to respond to if all other things are equal. Therefore, the last sentence gets a +0 rank, the second to the last gets a -4 rank, the third to the last gets a -8 rank, and so on.

3) Keyphrase Length Bonus: A long Keyphrase is generally a better match than a shorter one, so the length of the Keyphrase is added to the final Rank. If a * is used in the Keyphrase (See Advanced Section), then the Rank gets a -2, for an exact match is better than a spaced-out match. Other Keyphrase Plug-Ins have an adjusted Length Bonus.

Rank Advice: First, keep your ranks in a reasonable range. If you want a Keyphrase to be chosen above all others, then I would recommend giving it a Rank of 50. The highest possible Word Placement Bonus a Keyphrase can get is +12, and responses are generally no more than three sentences (-8 possible Sentence Order Bonus). In such a case the Bot will ignore all else and react to the Keyphrase ranked 30. Generally it's best to keep your Keyphrases Ranked between -10 and 50.


Emotional Impact
Keyphrases can be assigned an emotional impact. This emotional impact can be a value from -5 to 5, where the positive/negative aspect represents the kind of emotional impact (positive or negative) and the number represents the amount of emotional impact.

Using Emotional Impact: Bots have an emotional life- they remember things about you and form emotional relationships to you and others. The AI Engine handles recognition of most compliments and insults, but if you want your Bot to react emotionally to a wider range of statements, or give emotional impact to things that arent normally emotionally charged, just enter a number into the Emotion field that goes with that Keyphrase.

Range of Emotion: Emotional impact can range from -5 (very negative reaction) to 5 (very positive reaction). When triggered, the Bot will add this value to their overall emotional relationship to the one who sent the message.

Emotion Advice: Most of your Keyphrases should be kept at an emotional level of 0. Since the total emotional range of a Bot is from -5 to 5, it is best that when you decide to put an emotional reaction to a Keyphrase, that you keep your emotions values low, in the -1 to 1 range. Detected insults and compliments are generally given an emotional range of -1 to 1 as well. Create an optimist (positive-only emotional impact), pessimist (negative-only emotional impact), or moody bot (lots of varied emotional impact).


Responses
Editing Responses: To edit a Keyphrase and its Responses, click the Keyphrase. You can then create a number of Responses that are randomly selected from when that Keyphrase is found.

Randomization: Your Bot will remember which Responses have been said to which people, and will randomly choose a Response until they have all been chosen. Then it will clear the list and randomly begin again. This prevents the same Response from being given twice in a row, and spaces them out well enough to avoid repetition. For your part, have a good variety of Responses for each Keyphrase.

Once: The final column, marked with a "1" or the word "Once" indicates things that should only ever be said or asked once of each person. This includes a request for information that doesnt change from day to day, such as "Do you have any brothers or sisters?" Were I to ask you that more than once, you'd suspect something was amiss with my memory. At least one Response in each Keyphrase should not be marked "Once". NOTE: Generally it should be very rare that you should mark a Response "Once" (perhaps 1 in 100 Responses).

goto Keyphrase: This is referred to as a "Goto". Using "goto Keyphrase" causes a different Keyphrase's Responses to be selected. Note that if your only Response is"goto Keyphrase", it is often easier to just list the two Keyphrases together (See above).

<br> As this is all taking place in HTML, the standard <br> acts as a line break. This takes the place of hitting the "return" or "enter" button when typing in your responses.

Chapter 4: Special Keyphrases
There are a number of Keyphrases used only in special circumstances. Theseare central to any conversation and add a great degree of reality to the conversation. They are called "xKeyphrases" since each begins with an "x". They have strange titles to avoid them actually being found in a message and are used in the following circumstances:

Lulls & Nonsense
Special Personality Keyphrases: The following two Special Keyphrases are the most commonly used, especially when your Bot is relatively new. They can be central to your Bot's personality:

xnone: This is generally a large category used when no Keyphrases are recognized. Imagine that there is a lull in the conversation. This is your Bot's chance to bring up a new topic, ask a question, or whatever you like. If you wish to have an imaginary plot or series of events, this is where you can move the story forward (See Storyteller Bots).

xnonsense: This is used after a number of messages come across with no recognizable Keyphrases. Generally, responses of boredom or frustration appear here.

Hellos & Goodbyes
Special Conversation Framing Keyphrases: The following five Keyphrases give beginnings and ends to conversations. They are used when a conversation begins, ends, or is about to end:

xintroduce: This is used when a Bot initiates conversation with someone it has never spoken with before. It will generally contain an introduction of some sort.

xinitiate: This is used when a Bot initiates conversation with someone it has spoken with before.

xhello: This is used as a response to a new conversation initiated by someone else. Note that xhello will ONLY come up at the very beginning of a conversation. Repeated attempts to try to trigger xhello will trigger xnone instead, because it makes no sense conversationally to continue greeting someone again and again.

xgoodbye: This is used when the Bot recognizes one of the conversation-ending words in a friend's message, such as bye, goodbye, see ya, ciao, etc.

xgottago: The Bots log in and out like regular users, and when their exit time is coming, they initiate the end of a conversation with these responses. Responses here generally give a reason for having to go.

Favorites
Bots are able to automatically answer any question about what their favorites are, or which kind of thing they like the best. Following is the Keyphrase used for favorites:

xfavorite: These are Responses for questions that ask a Bot its favorite kind of thing.

Favorites Plug-Ins: There are the special plug-ins for favorites:

(type) This is the Type of thing the question is about. For "what's your favorite food" the Type is "food".

(specific) This inserts the chosen favorite of that type. It's good at knowing whether or not to include an article (a or an).

Commands
Bots can respond to Imperatives or "Commands" such as "Eat this pasta". In the Expert section, you'll learn how to make the Responses dependent on how the Bot feels about the person he or she is talking to.

xcommand: This is used when a Bot received a command such as "Put this lantern on the rock."

Plug-Ins: The following plug-in works in Responses to the xcommand Keyphrase:

(command) This inserts the command into the Response. For example, "I don't want to (command)."

Math
Bots can answer math questions. They can approximate an answer like a real person, give an exact answer, or refuse to answer, it's up to you. Following is the Keyphrase used for math:

xmath: When a Bot sense a math question, one of these responses are chosen.

Math Plug-Ins: There are the special plug-ins for math:

(answer) This gives an estimate of the answer if it is very large or very small, and the exact answer if the question is easy.

(exactanswer) This gives the exact answer no matter the difficulty of the question.