The study of wording and word interpretation for intermediate object on cooking instructional system with spoken dialogue

然而，在烹饪时阅读食谱有时是非常危险的，因为他/她在烹饪过程中必须拿着刀和火。因此，我们的目标是构建一个食谱指导系统，根据食谱文本指导如何烹饪。要实现这样的食谱指导系统，光看食谱是不够的。例如，食谱给出“Mix(1)to(2)”这样的说明是很常见的。这句话的意思是进程(1)给出的中间对象与进程(2)给出的中间对象混合在一起。如果系统只是用语音阅读这句话，厨师可能无法决定系统指定了哪些对象，因为他/她不知道/不记得进程(1)/(2)给出了哪个对象。在这种情况下，系统应该给出合理的名称，如“用前一道工序切洋葱”和“洋葱和胡萝卜的混合物”。这样合理的名字应该根据厨师对物品的理解而定。因此，我们提出了一种根据用户对烹饪过程中的中间食物对象的理解模型来合理命名的方法。在本研究中，我们提出了这样的中间对象倾向于由组件来指定，即，对象的餐名、器皿的名称、位置的名称、里程碑的名称(例如。“酱”，“面团”。我们提出了一种通过人类对中间对象的命名模型来指定指定的中间对象的方法。该模型保持了93%的对象指定准确率。根据人类对中间对象的命名模型，我们构建了一个基于实时视频烹饪通信软件IwaCam的食谱指导系统。IwaCam旨在实现远距离厨房空间之间的协作烹饪。

However cooking while reading a recipe is sometimes very danger because he/she has to handle a knife and fire in the cooking. Therefore, we aim to construct a recipe guidance system that instructs how to cook according to its recipe text. To realize such recipe guidance system, it is no enough to read the recipe book. For example, it is very common that a recipe gives such instruction as “mix (1) to (2).” This sentence means an intermediate object given by process (1) would be mixed with an intermediate object given by process (2). If the system just reads this sentence with a voice, a chef might not decide which objects the system designates because he/she does not know/remember which object is given by process (1)/(2). In this case, the system should give reasonable name such as “cut onion by the previous process” and “a mix of onion and carrot”. Such reasonable name should be according to like what the chef understand the object. Therefore, we propose a method to give reasonable name according to the model of user understanding toward an intermediate food object in a cooking process.In this research, we presented such intermediate objects is tend to be designated by components", "the name of meal the object going to be", "the name of vessel", "the name of position", and "the name of milestone (ex. "sauce", "dough). We proposed a method to specify the designated intermediate object by a model of human naming toward intermediate object. The proposed model archives 93% accuracy of object specification.According to the model of human naming toward intermediate object, we construct a recipe guidance system based on IwaCam, the a live video cooking communication software. IwaCam aims to enable cooperative cooking between distant kitchen spaces.