This work introduces the reader to a role-playing game based on dialogue systems implemented in AIML. Some aspects of natural language processing are presented, as well as examples of their applications, not only related to the game. Some aspects of semantic analysis and understanding of natural languages by a computer are described, however, the book is not a complete textbook about NLP.
AIML is presented as a specialised language for dialogue systems. It has very few options regarding memory of the bots. This work contains a description of implementation of our own method to provide time flow and impact of events on dialogues. Besides the description of the plot and setting, the publication also contains technical details and a manual for those who are willing to make their AIML bots reactive to some events. A complete working example is presented as well.
Several other applications of dialogue systems are presented, including call centres and website chatterbots. Prime Speech system for public transport in Warsaw is described as an example of automatic call-centre. Chatterbots are covered by Stanusch Technologies examples. Another application which is described is a game called Fa¸cde, which allows natural dialogues in communication with non-player characters.
A substantial part of the work is focused on a plot of Scarlett Mansion, which is an innovative role-playing computer game. The game is based on interactions (dialogues) between seven non-player characters involved in a crime. The difference to a standard solution is that a player can write anything to any of the characters and they respond accordingly, thanks to applied dialogue systems.
The theme of the game is that there were eight people in a house and one of them has been killed. A player is a detective who can speak with each of the other seven and by this conversation has to decide which of them is the murderer.
Some suggestions how to make attractive games are presented as well. These are keeping a plot in loops and providing challenges and victories in a good balance. In the end, plans for future development of Scarlett Mansion are described.
- Spis treści
-
1 Introduction 9
2 Dialogue systems 13
2.1 Meanings by compositionality 14
2.1.1 Semantics notation 15
2.1.2 Semantic analysis and modelling 17
2.1.3 Cyc and ontologies 19
2.2 Natural language understanding 19
2.2.1 Proper name identification 20
2.2.2 Part of speech tagging 21
2.2.3 Parsing 22
2.3 AIML 23
2.4 Modern trends in dialogue systems 26
3 Some examples of dialogue systems applications 29
3.1 Chattebots for websites 29
3.2 Automatic call-centre 32
3.3 Fa¸cade - Interactive drama 34
4 Scarlett Mansion plot 37 4.1 Setting 37
4.2 Well known facts 38
4.3 Facts known only by specific suspects 39
4.4 Crucial events 40
4.5 Reactions of suspects 41
4.6 Time limit 42
4.7 Elizabeth Scarlett 43
4.8 Colonel Mustard 45
4.9 Maria Scarlett 47
4.10 Eva Scarlett 50
4.11 Eric 52
4.12 Sir Garen Graf von Noxia und Demacia 55
4.13 Maid of the family (Miss Fortune) 57
5 Technical information 59 5.1 Bot states 59
5.2 Bot naming convention 60
5.3 Special sentences 61
5.4 Plot configuration file 62
5.4.1 Bot transitions 63
5.4.2 Crucial events 64
5.4.3 Time limit 66
5.4.4 Game end 67
5.5 Middleware 67
5.6 An example of a bot change 68
6 Attractions of role-playing games 73
6.1 Traditional dialogues 74
6.2 Game loops 75
6.3 Challenges 76
7 Plans and conclusions 77
8 Annex - Eva’s bot 85