Academia.eduAcademia.edu

Figure 6 – uploaded by Arnold Depickere

See full PDFdownloadDownload figure
Fig. 7. Frequency of AINI’s responses based on domain knowledge base.

Figure 7 Frequency of AINI’s responses based on domain knowledge base.

Related Figures (9)

Fig. 1. AINI and MSN authentication process.
Fig. 1. AINI and MSN authentication process.
Fig. 2. AINI and MSN messenger interface. Academic investigators and commercial developers have drawn into diverse research areas including natural language understand- ing (NLU), database systems, human factors and knowledge-based systems to make their systems more intuitive, user friendly and intelligent. Robert Planta and Stephen Murrell [44] investigate the development of a natural language (NL) interface for mixed ini- tiative dialogues within a constrained domain and demonstrates the applicability of the functional approach to Natural language system. In our paper, a significant difference between our proposed conversation bots is our domain matrix knowledge model as shown in Fig. 3. The domain knowledge model was aimed to play a major role in AINI systems. This domain model can be considered as a domain-dependent modular component and allows future improvements to encourage openness and collaborative contribu- tion to the specific knowledge domain. In our approach, we define the knowledge base of our conversation bots as a collection of spe- cific conversation domain units. Each unit handles a specific body of knowledge used during the conversation between the AINI and the IM buddies. For example, in the open-domain knowledge, the subject domains will cover subjects such as personality, business, We have outlined the conceptual and practical basis for the development of conversation bots for DesktopChat, WebChat and MobileChat which are supported by MSN Messenger protocol, as shown in Fig. 2. This will pave the way for a human-computer interface based on human natural language technologies. Hand- held devices provide an ideal platform for art and entertainment applications considering the growing number of mobile phone users world-wide. This will improve techniques for displaying con- tent, interaction, conversation and the emergence of wireless and shared interaction among networked users. Such applications can
Fig. 2. AINI and MSN messenger interface. Academic investigators and commercial developers have drawn into diverse research areas including natural language understand- ing (NLU), database systems, human factors and knowledge-based systems to make their systems more intuitive, user friendly and intelligent. Robert Planta and Stephen Murrell [44] investigate the development of a natural language (NL) interface for mixed ini- tiative dialogues within a constrained domain and demonstrates the applicability of the functional approach to Natural language system. In our paper, a significant difference between our proposed conversation bots is our domain matrix knowledge model as shown in Fig. 3. The domain knowledge model was aimed to play a major role in AINI systems. This domain model can be considered as a domain-dependent modular component and allows future improvements to encourage openness and collaborative contribu- tion to the specific knowledge domain. In our approach, we define the knowledge base of our conversation bots as a collection of spe- cific conversation domain units. Each unit handles a specific body of knowledge used during the conversation between the AINI and the IM buddies. For example, in the open-domain knowledge, the subject domains will cover subjects such as personality, business, We have outlined the conceptual and practical basis for the development of conversation bots for DesktopChat, WebChat and MobileChat which are supported by MSN Messenger protocol, as shown in Fig. 2. This will pave the way for a human-computer interface based on human natural language technologies. Hand- held devices provide an ideal platform for art and entertainment applications considering the growing number of mobile phone users world-wide. This will improve techniques for displaying con- tent, interaction, conversation and the emergence of wireless and shared interaction among networked users. Such applications can
Fig. 3. AINI’s Domain Matrix Knowledge Model.
Fig. 3. AINI’s Domain Matrix Knowledge Model.
Factoid question from TREC 8, 9 and 11
Factoid question from TREC 8, 9 and 11
Fig. 4. Top 10 Pagerank™ scale for bird flu domain.
Fig. 4. Top 10 Pagerank™ scale for bird flu domain.
AINI’s knowledge bases Table 2
AINI’s knowledge bases Table 2
Frequency of word from conversation logs Table 3 are quite small in comparison with images, video, or even docu- ments. These plain text messages, also known as instant messages, are the regular messages sent between principals on MSN Messen- ger. Sending a plain text message instructs clients receiving the message to display it on-screen, optionally with some simple formatting.
Frequency of word from conversation logs Table 3 are quite small in comparison with images, video, or even docu- ments. These plain text messages, also known as instant messages, are the regular messages sent between principals on MSN Messen- ger. Sending a plain text message instructs clients receiving the message to display it on-screen, optionally with some simple formatting.
Fig. 5. AINI conversation log with U00301 IM human user. In this section, some of the interaction features of the recorded chat are discussed and domain knowledge query will be our focus The excerpt shown in Fig. 5 is from a typical single session IM conversation exchanged between AINI and one of “her” buddies with ID U0025. This session illustrates the nature of the IM com- munication. Each new session will start with AINI being given a random greeting (message #1) eg. “Hi there:)”, “How are you today?”, “Hey:), “nice to meet u”. How I can call u?”, etc. These greetings will indirectly get some information about the user’s identity, such as their name or gender (utterance #2). In this ses- sion, we managed to identify U0025 user as ‘‘Hommer” and a “‘mal- e”. Although ‘“Hommer” used an emoticon to represent the first initial of his name ‘“(H)”, which refers also to “hot smile = F[x]”; however, AINI defined him as “Hommer’”. Following some dialogue (utterance #3), AINI gave a greeting to “Hommer” which came from open-domain knowledge. Since AINI’s knowledge is not equipped with full IM features such as acronyms, abbreviations and emoticons, utterance #5 shows that AINI failed to recognize the user input “sh F[x]”, which refers to “Same here” and “(smile)”
Fig. 5. AINI conversation log with U00301 IM human user. In this section, some of the interaction features of the recorded chat are discussed and domain knowledge query will be our focus The excerpt shown in Fig. 5 is from a typical single session IM conversation exchanged between AINI and one of “her” buddies with ID U0025. This session illustrates the nature of the IM com- munication. Each new session will start with AINI being given a random greeting (message #1) eg. “Hi there:)”, “How are you today?”, “Hey:), “nice to meet u”. How I can call u?”, etc. These greetings will indirectly get some information about the user’s identity, such as their name or gender (utterance #2). In this ses- sion, we managed to identify U0025 user as ‘‘Hommer” and a “‘mal- e”. Although ‘“Hommer” used an emoticon to represent the first initial of his name ‘“(H)”, which refers also to “hot smile = F[x]”; however, AINI defined him as “Hommer’”. Following some dialogue (utterance #3), AINI gave a greeting to “Hommer” which came from open-domain knowledge. Since AINI’s knowledge is not equipped with full IM features such as acronyms, abbreviations and emoticons, utterance #5 shows that AINI failed to recognize the user input “sh F[x]”, which refers to “Same here” and “(smile)”
Fig. 6. Frequency of dialogue topics. As discussed earlier, AINI’s domain knowledge model incorpo- rates several knowledge domains, thus merging the expertise of one or more experts. A ‘sales’ knowledge domain for instance, would contain expertise on improving sales, but it would also incorporate open-domain knowledge. Multiple domain knowledge, merged into AINI’s single domain knowledge would give the users the best conversation as show in Fig. 7.
Fig. 6. Frequency of dialogue topics. As discussed earlier, AINI’s domain knowledge model incorpo- rates several knowledge domains, thus merging the expertise of one or more experts. A ‘sales’ knowledge domain for instance, would contain expertise on improving sales, but it would also incorporate open-domain knowledge. Multiple domain knowledge, merged into AINI’s single domain knowledge would give the users the best conversation as show in Fig. 7.
Related topics:
Knowledge ModelingKnowledge ExtractionDomain Specificity
580 California St., Suite 400
San Francisco, CA, 94104
© 2025 Academia. All rights reserved