1

我正在写一些类似于smarterChild（那个时代的自动 AOL Instant Messenger 聊天程序）的东西，并且对我确定句子结构和响应的方法感到好奇。

目前，最基本的设计是使用 5 种主要方法，它们共同确定给定用户输入的适当响应：

    1) Read(): initially accepts user input, then calls first of many methods: sentenceType()
    2) getSentenceType(): User sentence type: Question, statement, directive, suggestion, etc…
    3) getWho(): If question is about the computer, or about someone else…
    4) getCategory(): What category the question is about (weather, sports…)
    5) getResponse(): finally, form a response

这些方法将查询数据库以确定用户输入是否包含关键字...

示例getSentenceType()：

public String getSentenceType(String sentence) {
    String type = null;

    for (String s : db.getQuestions()) {
        if (sentence.contains(s)) {
            type = "question";
            break;
        }
        else {
            type = "statement";
        }
    }
    return getWho(type, sentence);
}

返回句子结构的最终方法示例：

//final call... 
public String getResponse(String cat, String who, String type) {
    String response = new String();     
    String auxVerb, subject, mainVerb, noun; 

    auxVerb = "do";
    subject = who;
    mainVerb = "like";
    noun = cat;

    //question structure: auxiliary verb + subject + main verb + noun/pronoun
    //                      DO              YOU         LIKE        MARY?

    if (type.equals("question")) {
        response = subject + " " + auxVerb + " " + mainVerb + " " + noun +  ". "; 
    }
    else {
        response = auxVerb + " " + subject + " " + mainVerb + " " + noun  + "?";
    }

    return response;
}

示例数据库方法：

public String[] getQuestions() {
    String[] questions = new String[] {"why", "?"}; 
    return questions;
}

public String[] getWeather() {
    String[] weather = new String[] {"cold", "hot", "rainy", "weather"};    
    return weather;
}

然后它将逐步将所有方法结果连接成一个连贯的响应……然后将该结果发送回 Read()，它将结果打印给用户……

这是一种低效的方法吗？我知道，如果我继续走这条路……要建立一个强大的系统，将需要大量的if else检查和庞大的数据库来确定用户输入的每一种可能的响应类型。

有什么建议的方法吗？

谢谢

4

0 回答 0