1 Introduction

Confucianism is an Eastern ethical and philosophical system that has had great impact, especially in Asian countries, for more than 2,500 years. However, Confucian philosophies are extremely complex, thus may not be easily accessible to people, especially given the cultural and language barrier. Furthermore, traditional passive media provides only a linear understanding of his profound teachings, which restricts the learning process. In addition, young people are more inclined to use modern networked and social digital media. Hence, our vision is to enable users to experience this rich ancient culture using computing technology in an interactive and dynamic way.

Confucius Computer is a new form of illogical cultural computing based on the Eastern paradigms of balance and harmony, which is radically different from the ancient Greek logic normally experienced in computing. It aims to facilitate intergenerational cultural communication by enabling the young to gain deeper understanding of ancient Chinese culture using the modes of communication they are familiar with. The system uses new media to revive and model these historical philosophies and teachings, presenting them in new contexts, such as online social chat, music and food. This enables people to experience and explore ancient culture using the literacy of digital interactivity, and thus increases the physical and psychological proximity (Weaver 1997) of the older and younger generations.

The system core is a virtual Confucius thought engine that models Confucius knowledge from the Analects and his teaching method Yin Cai Shi Jiao, teaching student according to his or her aptitude (Hao 2007). Based on the model of Eastern mind and teaching, the system enables users to have meaningful social network communication with virtual Confucius. Based on the question asked or the sentence written, virtual Confucius identifies the Yu, relevant sub-domain of knowledge in the Analects. The system further determines the Hui, aptitude of the user on that topic based on the conversation history. According to Yu and Hui, Virtual Confucius then replies to the user. Users from our modern society could significantly benefit from this interactive and personalized advice from ancient virtual Confucius, which is not possible in passive media, such as the printed text.

The Confucius Computer system also offers a new artistic playground for interactive music-painting creation based on our Confucius music filter and the ancient model of Cycles of Balance. Confucius Computer also allows users to explore the traditional Chinese medicine concept of Yin-Yang through recipe mixing games.

Confucius Chat can be viewed as a new form of computer generated literature, which we call “Bit Literature”, an algorithmic composition of literature generated from both digital-bits and literature-bits. Bit Literature is very relevant in digital age because people’s interaction and understanding of literature is changing. We are in the age of instant messaging, short messaging system (SMS), Twitter, etc. and bits of literature may be better fitted in such communication channels. Bits of users’ input and bits of literature (small chunks of literature) from the knowledge database interplay with the computer algorithm to generate digital Bit Literature.

2 Motivation

Due to global aging population, issues of family intergenerational communication are becoming increasingly relevant (Gilford 1988). Confucius firmly believed that good family relationships were the key to reforming society. “The gentleman concerns himself with the root; and if the root is firmly planted, the Way grows.” (1:2, Analects). Family is important in developing one’s self-concept and its influence endures for a lifetime (Yerby et al. 1994). Grandparents play a vital role in a family, and the emotional attachments between grandparents and grandchildren are unique (Kornhaber 1981). Furthermore, people with more positive grandparent–grandchild relationships tend to have more positive attitudes toward older people (Silverstein and Parrott 1997; Soliz and Harwood 2003).

There have been encouraging results on the use of modern communication technology in bridging intergenerational gap, such as the use of video blogs for the old people to communicate with the young people (Gonzalez and Kurniawan 2008). Therefore, it motivates us to combine Confucius and Eastern philosophies and culture with the media literacy of the new digital generation through social networks, chat and interactive games. This novel form of illogical computing system, which is not based on Greek logic, allows young people around the world to experience deep Asian culture and values.

The system aims to increase the physical and psychological proximity (Weaver 1997) between the older and younger generations. Grandchildren are currently raised across the country or even on a different continent due to parents’ professional choices and family configuration, unlike previous times when most grandparents and grandchildren around the world stayed under one household (Nicholson and Zeece 2008). This results in lower physical interaction with their grandparents. Furthermore, with the rapid advancement of information and communication technology, there has been a decline of common media that allows both grandparents and grandchildren to interact. Also, media often portrayed negative stereotypes and under-represented characters of older people (Vasil and Wass 1993) and rarely promotes old culture.

Recent research on intergenerational communication across cultures has indicated that people in Asian nations construed older family and non-family members as less accommodating than did people in Western nations (Giles et al. 2003). It is possible that the lack of understanding of the grandparents’ culture, for example the ethic of filial piety, makes salient age-group identities and, thereby, triggers intergroup processes (Brewer 2003). However on the other hand, studies have also shown that strong traditional Confucian norms, such as filial piety and elders’ contribution to family harmony, have resulted in youths having more positive images of old age, closer psychological proximity and thus more respectful communication with older adults in the East (Giles et al. 2002; Levy and Langer 1994; Ng 1998). Thus, it is important that the young generation could interact with new media that promotes ancient philosophies and culture, so that they will have better understanding and communication with the older people.

3 Design steps

In order to understand the potential target user better, we adopted persona, a user-centered approach to create a personality profile during the system developing process (Pruitt and Adlin 2006). As we evaluate characteristics of the system design, well-crafted personas help us have a clear understanding of the probable users’ needs, thus allowing the technological ideas to fit in the real scenarios.

The personas describe the exposure and values toward culture. They are classified according to the engagement levels of the cultural knowledge and values. Ranging from an Outsider to a Believer, we have developed six personas as follows:

  • Outsiders People who do not care nor have much cultural sense,

  • Social Seekers People who get involved in culture-related activities mostly because of the social group they belong to,

  • Explorers Users who are interested in new and fun information and knowledge gathering. They often learn cultural knowledge when browsing,

  • Active Learners People who are highly interested in cultural knowledge and actively search for information on particular topics,

  • Searchers People who consider culture as a philosophy of life, often attend events and exercise in-depth learning,

  • Believers People who live with traditional culture and want to influence others.

The personas are summarized in Fig. 1. After the personas were constructed, three primary personas (Explorers, Active Learners and Believers) were selected as our main target user groups. According to the most prominent issues depicted within the primary personas, Explorers and Active Learners are more inclined to use modern networked and social digital media (e.g. internet), and are more likely to accept new knowledge. In addition, their attitudes toward cultural acceptance are positive, as compared to other groups, for example Outsiders, who may have high resistance toward accepting new cultural values. On the other hand, Believers, who constantly like to share their cultural values, may not have the right platform to communicate with the younger generations. Hence, Explorers, Active Learners and Believers are the most suitable user groups, as we are developing the system to promote cultural knowledge via computing technology.

Fig. 1
figure 1

Summary of the six personas we have developed

Full size image

Explorers are young children who like to learn new things and are interested in many kinds of fun knowledge or information, but they are restricted in learning cultural knowledge, as they have limited resources to those information. Active learners pursue cultural knowledge actively; they have basic sense of cultural knowledge within themselves, and consider it to be an important issue. However, there is no platform for them to do it naturally. Traditional culture is central to the lives of the Believers, and they think that the younger generations should embrace those values. However, their communication methods for giving advice, for example scolding and lecturing, are not well received by the younger generations.

We identified three scenarios which limit the cultural communications between the Believers, Explorers and Active learners. In the first scenario, we focus on communication of ethics or moral philosophy between the three groups of users. Explorers and Active learners use social network to communicate, and exchange ethical ideas and values but the Believers do not use social network. Confucius Chat bridges their communication by modeling the Believers’ set of ethical values and philosophies into a virtual chat agent, that the Explorers and Active learners could interact dynamically in a social network chat. In the second scenario, Explorers and Active learners enjoy listening to pop and rock music, while the Believers appreciate the traditional Chinese music, which promotes calmness and self-cultivation. Confucius Music provides a platform for the users to transform any music into traditional Chinese music, which gives them a new perspective to appreciate their favorite pop and rock music. The third scenario is concerned with food, which is essential to the intergenerational relationship. Having meals together over lunch and dinner promotes better ties between family members. The Believers would lecture the younger generations about the traditional Yin and Yang concept about food, when they are not eating the right combination of food. The children may not like to be lectured. Therefore, Confucius Food provides a fun platform for the Explorers and Active learners to learn traditional concept about food through recipe mixing games.

Figure 2 shows the relationship between Confucius Computer and the users. These new media allow the users to explore and experience rich ancient Eastern culture in an interactive and dynamic way. The platform we designed, in accordance with the target users key characteristics, will prompt them to interact with the system effectively and naturally. In addition, by drawing the culture closer to the users daily life, we made it easier for the users to engage in using our system. Therefore, they could better understand their older generations’ cultural values and philosophies, and in turn enhancing the intergenerational communication.

Fig. 2
figure 2

Relationship between Confucius Computer and the users

Full size image

4 System description

The following subsections elaborate on the Confucius Computer system that has the Confucius Chat, Confucius Music-painting and Confucius Food components. Confucius Chat introduces a social network chat agent, which models Confucius teachings, thus enabling users to have meaningful social network communication with virtual Confucius. Confucius Music-painting offers a new artistic playground for interactive music-painting creation based on Confucius music filter and the ancient model of Cycles of Balance. On the other hand, Confucius Food allows users to explore the traditional Chinese medicine concept of Yin-Yang through recipe-mixing games.

4.1 Confucius Chat

Confucius Chat offers an alternative approach to existing artificial intelligence (AI) chat agents, for example ELIZA, ALICE, PARRY, TINYMUD and Hex. These chat agents use simple pattern matching approaches and employ tricks to cover-up the failure of understanding the user input (Sing et al. 2006). The tricks could be frequently switching topics, or rephrasing the input by replacing the first person’s pronouns to second person’s pronouns and vice versa. These approaches fail to help users gain further understanding on the topic of discussion, thus offer minimal benefits to the interaction. Furthermore, existing chat systems provide logical and linear responses (Sing et al. 2006). On the other hand, our system is based on Eastern Confucius logic to produce non-linear and meaningful illogical responses. Table 1 summarized the performance comparison between Confucius Chat and other popular chat agents.

Table 1 Performance comparison of various chat agents
Full size table

In modeling Eastern logic and Confucius philosophies, our chat system employs extensive advanced information retrieval techniques and natural language processing techniques. First, the Confucius Analects teachings in the database underwent a natural language processing engine to identify the Yu, relevant sub-domain of knowledge conveyed by each sentence. This includes topic and emotion detection engine, Part-of-Speech (POS) tagging (Kupiec 1992) engine, sentence similarity measurement engine and verb frames detection engine. In this process, entries in the Analects are broken down into small bits of literature. The user’s input is also processed in a similar manner. The key information from the user’s input is then compared with the Yu, for the best match according to Leacock Chodorow (LCH) similarity measurement model (Leacock and Chodorow 1998), built according to the taxonomy provided by WordNet (Fellbaum 1998).

Furthermore, with proper inclusion of the information identified in the earlier chat entries, virtual Confucius determines the Hui, aptitude of the users on the Yu, and generates a suitable reply. This method also ensures the continuity of the conversation with the user concurrently. Finally, the dialogs are also added to the knowledge database, to expand the database entries (bits of literature) and the induction rules in the reply generation process.

Figure 3 shows the technical block diagram of the entire Confucius Chat system. The details of the algorithm will now be explained. Among the three processing engines employed, topic and emotion detection engine relies on a topic vector space of n dimensions. The vector space is constructed using n topic as n dimension, and the n topics are chosen according to the collection of topics from the Analects. Subsequently, every sentences is measured against all n topics and assigned the likelihood of it belonging to each topic X i where i  ∈ (1, n). Subsequently, the sentence is positioned in the vector space denoted by vector V sentence ,

$$ \hbox{V}_{\bf sentence} = [X_1, X_2 {\ldots} X_n]^T $$
(1)
Fig. 3
figure 3

Algorithmic flow diagram of the Confucius Chat system

Full size image

We apply the same method and position the user’s input sentence in the same vector space, and obtain the vector V Analect , then, we can calculate the projection Pr of V Analect onto V input using the vector dot product.

$$ \hbox{Pr} = \hbox{V}_{\bf input} \cdot \hbox{V}_{\bf Analect} $$
(2)

This scalar projection Pr is the measurement of likelihood with respect to all possible topics that can interlink the input and the Analects sentences in the database. The emotion detection method is similar to the topic detection except that we make use of Russell dimension for emotions (Russell 1980) that classifies the emotions into two dimensions, the excitement-calm axis and the pleasant-displeasing axis. The POS tagging engine employs standard Hidden Markov Model tagger, which has gone through supervised training of a large corpora to enhance the precision of tagging. The sentence similarity engine employs the LCH word similarity measurement, which is based on the location of the two words in the selected taxonomy tree structure. The similarity score is given in (3), where W 1, W 2 are the words being measured, and depth (W) is a function that measures the depth of phrase W in the single rooted WordNet taxonomy tree. W LCS is the least common subsummer (LCS) (Wu and Palmer 1984) of W 1 and W 2.

$$ \hbox{Similarity}(W_{1},W_{2})= \frac{2 \times \hbox{depth}(W_{\rm LCS})}{\hbox{depth}(W_{1})+\hbox{depth}(W_{2})} $$
(3)

With the score between every word in sentence S 1 and sentence S 2, we can calculate the weighted average of the two sentences based on (4), where W i is the enumeration of words in S 1, and W j is the enumeration of words in S 2.

$$ \hbox{Similarity}(S_1,S_2) = \frac{\sum_{i=1}^m \sum_{j=1}^n \hbox{Similarity}(W_i , W_j)}{m \times n} $$
(4)

The bits of literature are then generated from a set of the most relevant candidate words or phrases from the Analects entries and a new sentence is composed according to a wealthy list of pre-defined sentence structures, ranging from traditional Chinese sentence structure to modern English sentence structure. Alternatively, it could dynamically generate sentence structures with distinct styles with the information stored in the knowledge base previously summarized by scanning through various collections of literature tagged with language style. At this point, natural language processing techniques are again employed to ensure the correctness of the sentence grammatically and semantically. Virtual Confucius bit literature generation is personalized and would benefit the user directly. During the chat interaction, the intermediate results with intuitive representation of the bits literature generation are shown dynamically for the user to gain insight into the modeling of virtual Confucius thoughts.

A screenshot of the chat history with virtual Confucius in MSN is shown in Fig. 4. To explain one example, when the user asks, “I really do not understand what is the purpose of love in our life.”, the system first tokenizes the sentence into words. The system then finds synonyms for those words. Following that, the system determines the relative weight of information among those words based their inverse term frequency (Jones 2004). In this case, the highest weighted words and their weights are are shown in Table 2.

Fig. 4
figure 4

Chat history with virtual Confucius in MSN

Full size image
Table 2 The highest weighted words and their weights
Full size table

The algorithm will then select several Analects sentences that have the closest proximity, based on (4), to the those highest weighted words and their synonyms. The selected Analects sentences are shown in Table 3.

Table 3 Selected Analects which have the closest proximity with the user’s input
Full size table

POS analysis is done on the selected Analects sentences words and their synonyms, which have high weight of information, to generate a list of candidate words as shown in Table 4. Similarity, emotion and importance of words are compared between the selected input and Analects words. The words with the highest scores are then filled into the manually crafted sentence structures stored in the database. Virtual Confucius then replies “Not delight to love, not love for virtuous aim”. As we can see, the output sentence is non-linear (not direct input-output pattern matching) and has the surprising element that makes the user think deeply about his question.

Table 4 Selected words from the Analects sentences
Full size table

4.2 Confucius music-painting

The Confucian interest in music is due essentially to its educational and ethical motivations; and the primary role of music is to build one’s moral character. Confucius Music introduces algorithms to filter and transform any type of music into “positive” music by fusing Confucius philosophy with music theory. This “positive” music is believed to promote personal character development (Zheng 2004) as well as heal mental and emotional state of a person (Cook 2002).

For centuries, music has been used to heal both the mind and the body (Palakanis et al. 1994). Research has shown that music, when delivered properly, will have immediate effects for example greater mental acuity, deep relaxation and even help to cure insomnia (Goldman and Jonas 2006). One common example is to utilize the concept of consonance to the patient, where two tuning forks with different frequencies are played at each ear at the same time (Munch 2005). Studies have also been done to prove that music have many effects on one’s physical changes including pulse rate change, blood pressure, digestive, respiration rate, electrical skin resistance, activity, and electrical activity in the brain (Munch 2005). For instances, music with less insistent rhythm helps to make listeners feel more calmed and relaxed, thus increases the ability to absorb more information (Dreikurs 1960).

Confucius music filters the timbre, rhythms and scale of the input music into “positive” music. Confucius music filtering is achieved by passing the original music Musical Instrument Digital Interface (MIDI) file through a combination of three filters, namely Pentatonic filter, Gradient filter and Bezier filter.

Based on music theory, pentatonic scale contains the least amount of dissonance (Kolinski 1962). Dissonance in music is associated with disharmony that causes intense emotional feelings to the listener. To use the Pentatonic filter, the user has to choose a desired root note for the Chinese pentatonic scale that corresponds to the cosmological theory of the five elements: metal, wood, water, fire and earth (Huang 1963), and the system will construct an array containing the members of the scale. The incoming notes that do not match any of the members in the array will be transformed to the nearest member.

Confucius emphasized that music should be balanced, evident in the Analects records “The Guan Ju expresses joy but now debauchery, sorrow but not morbidity” (3:20, Analects). Gradient filter improves the listeners’ mental state and stimulate calmness as well as relaxation by smoothing the drastic changes of pitch in the music and output a balanced music. To do this, the gradient \(\left(\frac{\hbox{pitch}\,\hbox{difference}}{\hbox{time} \,\hbox{difference}}\right),\) shown as red line in Fig. 5, is calculated and analyzed. In this example, the gradient between the notes A and B is \(\left(\frac{y2 - y1}{t2 - t1}\right),\) where y2, y1 are the MIDI note numbers that correspond to the pitch of the notes, and t2, t1 are the time when the notes are played.

Fig. 5
figure 5

The horizontal and vertical continuous lines represent the time and pitch differences respectively and the dotted lines show the gradient. y1 and y2 are the pitch values of the notes A and B respectively, while t1 and t2 are the times when the notes A and B are played respectively

Full size image

When the gradient deviates from the average by p%Footnote 1, the pitch and time differences are analyzed and brought down to the ceiling and floor respectively. The ceiling makes sure that the pitch difference will not exceed the average by more than p%, and the floor makes sure that the time difference will not be shorter than the average by less than p%. We tested the music filter with varying p value, and noted that if p is too large, no significant shifting will be done, but if p is too small, most of the notes are shifted down to the ceiling, and time lengthened by the same amount, causing the song to be monotonous. Through experimentation, p is chosen to be 50. Figure 6 shows the song in Fig. 5 after the gradient filter. The circled note has been shifted down, since originally it has very high gradient.

Fig. 6
figure 6

Song after Gradient transformation

Full size image

Bezier filter uses a more complex method to create a smoother transition between notes and at the same time retains the musical structure by combining averaging method with Bezier curve approximations (Angel 2009). The first step involves similar method of analyzing the time difference between the notes, and making sure that the time difference will not be shorter than 50% below average. Then, by using Bezier formula as stated in (5), the filter creates bezier points based on degree and smoothness \((\hbox{subdivision} =\frac{1}{\hbox{smoothness}})\) parameters. The original musical notes are then replaced by the Bazier points corresponding to the same time axis.

$$ Q(t) = \sum\limits_{i=0}^n \,(^{n}_{i})(1-t)^{n-i}t^{i}P_{i} $$
(5)

where

  • Q(t) is the Bezier point for different values of t;

  • n is the degree of curve (number of control points-1);

  • t is the subdivision ratio that ranges from 0 to t;

  • P i is the value of the control point.

In our experimentation, observations are noted that at high order/degree and smoothness value, computation becomes very expensive, thus leading to slow performance. Low order/degree and smoothness value does not cause any significant smoothing. We have set the order/degree in the range of 20–100 and smoothness value in the range of 20–100 with increment of 10. Figure 7 shows the song in Fig. 5 after the Bezier filter. Two notes have been shifted down, and one note has been shifted up (circled).

Fig. 7
figure 7

Song after Bezier transformation

Full size image

Further analysis is done to tackle the dissonances that may be created as a result of the transformation. The ratio of the multiple notes are checked against a list of consonance ratios (Schellenberg and Trehub 1994) and if it does not belong to the list, one or more of the notes will be shifted so as to eliminate any dissonances. This step does not affect the musical structure significantly as the note will not be shifted much in the scale.

The users also experience Confucius’s philosophy about painting, which is to aid moral cultivation (Ortiz 2009), and at the same time learn about the ancient Chinese model of Cycles of Balance and the relationship between the five elements. Output of the music filter is visualized in the form of traditional Chinese painting. Mapping between the five elements and the musical notes are based on the ancient Chinese theory concerning the form of the elements (Nakaseko 1957). For example, earth corresponds to the root note because it has the most definite form. Water corresponds to the last note of the scale because it is formless.

The system allows bidirectional music-painting interaction that enables users to manipulate the Chinese painting to generate music. In the interaction mode, as shown in Fig. 8, the user can scroll through the painting, drag and drop the existing painting objects, or add in new objects. When a new object is added, it would generate a reaction with the adjacent painting object based on the Cycles of Balance as shown in Fig. 9. In an example in Fig. 10, the user dragged a wood object to a fire object and the fire object became larger, as wood generates fire.

Fig. 8
figure 8

Confucius Music-painting interaction mode screenshot. On the top left corner, user can click to drag any of the five elements to the music scroll in the middle of the screen. The music scroll’s horizontal axis corresponds to the time and the vertical axis corresponds to the pitch. The user can drag any painting objects to a new position on the scroll, which will change the time the note is played and it’s pitch. The cycles of balance icon on the top right corner shows animation of generation and destruction of elements when activated by the user

Full size image
Fig. 9
figure 9

Cycles of balance and the relationship between musical notes and the five elements

Full size image
Fig. 10
figure 10

Example of the generation of fire, based on the cycles of balance, in the interaction mode

Full size image

Upon finalizing the painting arrangement in the interaction mode, the corresponding output music is transformed into a dynamic chinese painting. A cycles of balance truth table, as shown in Table 5, is used to describe the mapping relationship between the pentatonic notes and the generation and destruction of the five elements, based on the current and the previous music notes. We employ a cyclical algorithm based on the “When, Where, Who, What, Why and How” (5W1H) thinking strategy (Ng and Tan 2004) that generates the story elements which consists of when does it happen, where does it happen, who are involved, what happened, why does it happen and how does it happen. Why does the event happen is explained by the Cycles of Balance. How does the event happen is explained in Table 6.

Table 5 Cycles of balance truth table
Full size table
Table 6 Creation and destruction events
Full size table

The creation and destruction events in Table 6 are based on traditional Chinese culture. For example, in Chinese culture, swallow is a messenger of spring, it brings happiness and blessings. Lotus is a symbol of summer, whereas plum flower is a symbol of winter. Five human relations are based on Chinese folk stories and classical works. For example, for describing husband–wife relationship, we employ well-known ancient Chinese love story in “A Dream of Red Mansions” (Xueqin and Xianyi 1999).

A screenshot of the painting is shown in Fig. 11. In this example, a series of notes AGAEEG were played. The note G played after note A, triggered the summer theme which corresponds to “when” in the narrative. Lotus, which symbolizes summer in Chinese painting is displayed. However, due to the destruction action, the lotus and crane on the painting disappeared after one second. The following note A did not trigger any event as there is no relation between note G to note A. Subsequently when the note E is played, the system triggered a creation of grass land event which corresponds to “where” in the narrative. The following note E did not trigger any event, as there is no relation between note E to note E. The next note G triggered the event of parent–children that corresponds to “who and what” in the narrative. In this series of notes, a narrative of parent and children playing together on the grass land in the summer were generated in the painting.

Fig. 11
figure 11

Screenshot of the final chinese painting generated from Confucius music-painting system

Full size image

4.3 Confucius food

Confucius emphasized that being physically healthy is an act of filial piety. According to traditional Chinese medicine, the human body is a miniature universe (Rong et al. 2007). To maintain a healthy body is to maintain a balance of Yin-Yang. One way to achieve this is by choosing the correct food which is divided into hot, cold and neutral. Factors that influence the choice include the current body state and the external environment (e.g. season). Using deep modeling of such philosophy, the system allows young people to gain insights into the complex concept of Yin-Yang in a unique context of food through recipe-mixing games. Recently, food-related games have gained popularity among young people. We aim to use new media entertainment way to motivate the young people to learn traditional Chinese concept about food.

It was noted in the book Feeding China’s little Emperors (Jing 2000) that in China, young children have already lost touch with traditional food concepts. Using deep modeling of Traditional Chinese Medicine (TCM) concepts, we have created an expert system (Ignizio 1991) that users can interact with to gain insights into the complex relationship of Yin and Yang in the unique context of food. The system has two mini-game prototypes involving cultural festivals to allow users to experience this TCM concept in a fun and engaging way. The two festivals are the Mid-Autumn and Thanksgiving festivals. Before the start of the game, virtual Confucius queries the user’s body state by prompting the user to answer some questions typically asked by traditional Chinese physicians. In the first game, the user first has to choose the ‘skin’ or crust of the mooncake followed by up to three fillings. The system calculates the Yin and Yang value of the mooncake. This value is then compared with the user’s personal body state to determine if the mooncake is suitable for her. It offers recommendations and alternatives if the mooncake is unsuitable, as shown in Fig. 12. The game for the Thanksgiving festival is similar except a stuffed turkey is created instead.

Fig. 12
figure 12

Confucius Food recommendation screenshot. The user has a hot body state, and he selected Suzhou style mooncake skin, deer and ginseng which are hot food. Confucius advised the user that the food is too hot for her, and recommended some cold food, for example bitter gourd, snow fungus and banana

Full size image

We have extended the existing system to include modern health science in its knowledge base. The new system permits users to compare and contrast the different health systems in the context of dietary planning. The system then takes into account factors from both TCM and modern health science when considering its recommendations. For example if the meal is too ‘hot’ for the user based on TCM, and also too high in calories based on modern food science, the system will search for ‘cold’, low calorie foods.

5 User study

5.1 Users lifestyle study

We have conducted a study with 6 pairs of Singapore Chinese parent–child (12 users) to better understand their usage patterns of digital media and understanding of Confucius and ancient Chinese philosophies. The children are aged 7–11, while the parents are aged 38–52. At this stage, the users have no prior experience interacting with Confucius Computer system. They were asked to fill up a questionnaire. The results are summarized in Table 7.

Table 7 Children and parent’s feedback before using Confucius Computer
Full size table

From the preliminary questionnaires, we noted that both parents and children play digital and traditional games. The children played an average of 6.83 h of digital games per week, when compared to 2 h per week for traditional games. Digital games they played include windows spider, educational titles, racing game, Maple Sea, Grandchase, Cooking mama, Diner dash, Maple story and Boogie. Traditional games they played include hopscotch, skipping rope, Monopoly, UNO, chess, various forms of board games and card games. The parents played an average of 3.5 h of digital games per week, when compared to 1.5 h per week for traditional games. Digital games they played include Windows spider, Freecells, Solitaire, educational titles, Yahoo games, Zuma, AtlanTis and MagicMatch. Traditional games they played include Monopoly, UNO, chess, various forms of card games and board games. It is notable that the children spent more time on games when compared to their parents, and in both groups they spent more time on digital games when compared to traditional games. Most of the children and parents are using social networking applications like Facebook, MSN, Google chat. Some of the children also play massively multiplayer only role-playing game MapleSea.

100% of the parents agreed that they are able to understand young people’s way of making sense of things, while only 50% of the children agreed that they are able to understand old people’s way of making sense of things. The children gave low rating for understanding of Confucius philosophies and ancient Chinese culture, while the parents rated themselves neutral or high in understanding. 50% of the children agreed that knowing about ancient culture and philosophy is fun, while 100% of the parents agreed to this statement. None of the children uses digital media to find out about ancient culture and philosophy. Few parents reported using online search engine, for instance Google and Baidu, to find out more about ancient culture and philosophy.

5.2 Preliminary user study

We would like to evaluate the usability and functionality of the systems before engaging users of different generations. Two studies were conducted for Confucius Chat and Confucius Music-painting, as elaborated in the subsections below. Through these studies, we seek to understand the users’ comprehension and usage of the systems, the systems’ functionalities and the user experience.

5.2.1 Confucius Chat study

A user study was carried out with 14 university students. Of the users, 8 are ethnic Chinese, 5 are Sri Lankans and 1 Iranian. The average age of the users is 26.21. The users text chatted with virtual Confucius through a Java applet in their web browsers. Their chat histories were recorded. After the chat session, they filled up a questionnaire, and the results are shown in Table 8.

Table 8 User study results from with 14 university students, after interacting with Confucius Chat
Full size table

Question 1 gauged the user’s enjoyment in interacting with virtual Confucius. 50% of the users agreed and strongly agreed that they enjoyed interacting with virtual Confucius, 35.7% of the users are neutral, and only 14.3% of the users disagreed that they enjoyed the interaction. For second question, “Virtual Confucius’s reply is related to the user input”, 50% of the users agreed and strongly agreed, 28.6% of the users are neutral, and only 7.1% of the users disagreed. Upon looking into the chat history, we noticed that the users who disagreed that the chat output is relevant to their input, used many short sentences of only two or three words. The system performs worse in this condition as these input sentences contain no keywords that could trigger meaningful topics. Of the users, 64.3% agreed to question 3 that virtual Confucius’s reply is interesting.

Results from questions 4, 5 and 6 suggest that in the opinion of most users, traditional philosophy, for example Confucius’s teachings, is demolishing in the new generation, and they believed that Confucius Chat could help to preserve and promote Confucius’s teachings. It is interesting to note that more users agreed and strongly agreed that our system could promote Confucius’s teachings, when compared to only preserving his teachings. In question 7, 64.3% of the users agreed and strongly agreed that our system could be used for cultural communication, while 35.7% of the users are neutral with this statement. In question 8, the users gave overwhelmingly positive response that they like to explore culture and philosophy through social network chat. Of the users, 92.9% agreed and strongly agreed with that question. In question 10, more than half of the users prefer to talk to virtual Confucius, when compared to reading philosophy books.

5.2.2 Confucius music study

A study was carried out with 13 participants, aged 20–37, to evaluate Confucius Music. The participants are all Asians, which comprises 7 Chinese, 3 Sri Lankans, 1 Malaysian, 1 Burmese and 1 Vietnamese. Three songs have been filtered through 7 different combination of filters resulting in a total of 3 different groups, each having 8 different songs (original and 7 filtered songs). There were a total of 24 songs for this study. The songs were played to the users in random order and no information about the songs was given to the users. This was done to prevent the users from making any pre-judgment about the songs, and hence affect their responses to the study questions. Also, all the songs were played using harp software synthesizer to neutralize the effect of the different instrument sound on the perception of music.

After each song was played, the users were asked to rate from a scale of 1 to 10 (10 being most Chinese like), whether the song sounded Chinese like. Also, they were asked to rate their feelings from 1 to 10 (1 being excited and 10 being calmed).

Results in Table 9 suggest that songs that were filtered through pentatonic filter and other pentatonic combination filters sounded more Chinese like. The average user rating for the original songs is 3.49. Songs that passed through non-pentatonic combination filters has an average rating of 3.71, while songs that passed through pentatonic combination filters has an average rating of 6.09.

Table 9 Results on how Chinese-like the songs are
Full size table

As we look further into the feelings of the users after listening to the song, we noticed that songs that passed through combinations of the Gradient, Bezier and Pentatonic filters make them feel slightly more calmed when compared to the original song. Table 10 shows the overall results (average value of all the 3 songs) of the users’ feelings after listening to the songs. We have verified this result with users’ verbal feedback that the music made them feel more calm. Some users’ comments are shown in Table 11.

Table 10 Feelings of users after listening to the songs
Full size table
Table 11 Users’ comments about the filtered songs
Full size table

5.3 Intergenerational study

The same users from the Sect. 5.1 were engaged for the intergenerational study. The users were given a short introduction about Confucius Computer and shown the concept video. They were then shown how to use the systems and had their first try. At this point of time, the facilitators helped the users to familiarize themselves with the user interface and tried out various functions of the Confucius Computer system. Each user then played in three settings, alone, with peers, and with another parent or child (intergeneration). They were asked to fill up a questionnaire after playing in three settings.

The results of the study are summarized in Table 12. Questions 1, 2 and 3 seek to evaluate the user’s enjoyment of interacting with Confucius Chat, Confucius Music-painting and Confucius Food, respectively. The users gave overwhelmingly positive response to their experience of interacting with our systems. Of the children, 83.3% agreed that they enjoyed the chat and music-painting applications, while 66.7% of them agreed that they enjoyed the food application. Of the parents, 83.3% agreed that they enjoyed the chat application, while 100% of them agreed that they enjoyed the music-painting and food applications.

Table 12 Children and parent’s feedback after using Confucius Computer
Full size table

Questions 4, 5, and 6 seek to understand the user’s enjoyment of interacting with the systems alone, with their peer and with another younger or older user (intergeneration). Fifty percent of the parents either neutral or disagreed that they enjoyed using the systems alone, while 66.7% of the children disagreed that they enjoyed using the systems alone. Most of the users preferred to play Confucius Computer together with another person. Of the parents, 100% agreed that using Confucius Computer with their children was enjoyable, and 50% of them agreed that using the system with their spouse was enjoyable. Of the children, 83.3% agreed that they enjoyed using the system with their peers and 66.7% of them agreed that they enjoyed using the system with their parents. However, only 33.3% of the children agreed that using the system alone is enjoyable.

Questions 7, 8 and 9 seek to evaluate the user’s perception of the systems in improving intergenerational interaction and their readiness to recommend this media to their friends and family members. Hundred percent of the parents and 83.3% of the children agreed that Confucius Computer can be used to improve interaction among young and old people. Of the children, 100% would recommend this media to their friends and family members, while 66.6% of the parents would recommend this media to their friends and 50% of them would recommend this media to their family members. Rationale for lower rating from parents with regard to recommending this media to the children was due to their concern that the philosophies and teachings are too complex and younger children may not comprehend them. However, the children did not have such concern and they were excited about their interactions with Confucius Computer.

After the questionnaires, we conducted a focus group session where the users were asked to comment about Confucius Computer. Some of their feedbacks were shown in Table 13. The children suggested making the system online and they look forward to play new Confucius Computer applications. One parent suggested making the system available not only on websites, but also on mobile phones through short messaging system. The chat system is currently available from our website (http://www.confucius.mixedrealitylab.org) and on MSN. The applications are made into Java applets which can be easily hosted on websites and mobile phones.

Table 13 Users’ comments about Confucius Computer
Full size table

6 Conclusion

In this research, we have created a novel merging of ancient culture with recent media literacy to mediate intergenerational communication. Our studies showed that users gave positive feedbacks to their experience of interacting with Confucius Computer. They enjoyed using Confucius Computer and would highly recommend this media to their family members and friends. They believed that this media could improve intergenerational interaction. The user studies results gave us insights into the intergenerational user’s experience. We would like to conduct more in-depths user studies to understand the impact of cultural media, like Confucius Computer, in mediating intergenerational communication. We hope this work will in future be used to allow new interactive experiences with all forms of deep traditional culture, including Greek, Arabic, African culture.