인공지능(AI)은 20년 이상 게임 분야에 널리 적용되어 왔다. 그러나 협동(coordination) 게임에서의 AI 에이전트, 특히 경주 게임에서 협동에 대한 연구는 상대적으로 적은 주목을 받아왔다. 이러한 관심의 부족은 불완전한 파트너를 충분히 보완하면서 사용자의 게임 플레이 경험 과 수행 능력 을 저해하지 않아야 하는 복잡성에서 부분적으로 기인한다. 우리는 경주 게임에서 협동 에이전트 의 잠재력을 탐구하고 밝히기 위해, 자동차 컨트롤을 두 개의 서로 다른 에이전트로 나눔으로써 협동 환경을 갖춘 자동차 경주 게임을 개발하였다. 이어서 실험을 통해 다양한 훈련 방법과 파트 너의 정보를 활용하여 에이전트와 파트너의 협동을 평가하였다. 특히, 학습 시 서브-옵티멀 파트 너와 함께하는 것과 에이전트를 해당 파트너에게 맞게 개인화하는 것의 영향을 조사하였다. 연구 결과, 불완전한 파트너와 훈련했을 때 성능이 2%에서 7%까지 향상되었으며, 파트너에게 맞게 개 인화했을 때는 모든 파트너에게 일반화한 경우보다 최대 3점(6.7%)까지 성능이 향상하였다. 본 연구를 통해, AI 에이전트를 개인화하는 것의 잠재력을 보여주었고, 에이전트가 파트너의 불완전 함을 인지하는 것의 장점을 확인하였다. 본 연구가 협 동 게임에서 개인화된 에이전트 연구에 이 바지하기를 기대한다.

Fast-paced advancements in technology demand swift adaptation and presents new opportunities and challenges for the optimization of communication, especially for advertisers. Digitalization and new developments in ICT have brought significant changes to the ways in which information, especially promotional messages, is disseminated to consumers. Additionally, with explosive interests in anticipation of fully autonomous vehicles, this study identifies and addresses the potential to optimize communication in an under examined digital media environment – in-vehicle infotainment system. Therefore, this study proposes a text-image embedding method recommender system for the personalization of multimedia contents and advertisements for in-vehicle infotainment systems. Unlike most previous research, which focuses on textual-only or image-only analyses, the current study explores the understanding, development and application of text embedding models and image feature extraction methods simultaneously in the context of target advertisement research. Overall, this study highlights the need to adapt to the ever-evolving technological landscape to optimize communication in various digital media environments. With the proposed text-image embedding method, this study offers a unique approach to personalizing multimedia content and advertisements in the under-explored digital media environment of in-vehicle infotainment systems.

This research intends to examine whether these types of data – behavioral versus social networking data – affect consumer response to personalized ads. With the advancement in technology, marketers have access to various types of personal data, including their online/offline behaviors and social networking activities, and use those data to retarget consumers. Moreover, this research examines the moderating role of SNS privacy concerns on consumer response to different types of retargeted ads. The findings of this research may offer theoretical and practical implications to understand consumers' responses to retargeted ads that use social networking information.

Consumers may not be fully aware of the level of personalization used in brand placements in digital advertising. Personalized brand placements use customer data to select and insert preferred brands into digital media content, such as having Coke appear in a YouTube video for someone who favors Coca-Cola. This personalization can enhance the consumer experience (Trifts & Aghakhani, 2018) but may also backfire if placements don't account for changes in brand preference or when personalized placements are co-consumed with a person whose preferences strongly diverge from one’s own (e.g., one’s romantic partner; Brick, Fitzsimons, Chartrand, & Fitzsimons, 2018). Brands have a significant impact on how people perceive each other. Brands convey social information and symbolic meaning (Keller, 1993; Levy, 1959) and can influence behavior (e.g., Fitzsimons, Chartrand, & Fitzsimons 2008) and provide a source of affiliation (Escalas & Bettman, 2003). However, less is known about how brands affect basic social connections like empathy. This notable gap is relevant to the evaluation of the impact of brand placements in digital advertising as brand logos are quite visible in brand placements and can be the first piece of information someone observes when forming an attitude towards another person. For example, imagine an Instagram post of a person wearing a Harley Davidson baseball cap. What kind of person did you imagine? Does your impression change if, instead, the brand was People for the Ethical Treatment of Animals, or the Trump Organization? Moreover, how likely would you be to engage with media showcasing a person supporting this brand. In the current research, we propose that personally disliked brands create a negative bias in person perception and consumer judgment, a phenomenon we coin “brand negativity bias.” Our results show that disliking a brand can reduce empathy for others and substantially bias attitudes toward a digital product associated with but not created by the brand. Specifically, we show process evidence that suggests disliked brand placements reduce (1) empathy toward others through reductions in perceived similarity and (2) attitudes toward a digital product through perceived similarity and empathy, as serial processes. In addition, although we do not empirically test an overall mechanism that explains the negativity bias in its entirety, we speculate several reasons that underlie the broad influence of personally disliked brands (e.g., symbolic, ideological, or moral). From a societal point of view, our findings inform the public to be mindful of how simple consumption cues can bias people’s decision making, which holds implications beyond consumer judgment in domains involving first impressions.

Personalizing banner ads or embedding ads with specific data signals or triggers, such as – personal characteristics, past behaviours, etc. is believed to improve customer response or Click Through Rate (CTR) since, embedding ads with viewers/recipients’ personal data or characteristics make ads more appealing and relevant to users (Lambrecht & Tucker, 2013). However, evidence also exists in literature that personalization can be ineffective as the usage of customers’ personal data can trigger off privacy concerns causing them to ignore such ads (van Doorn & Hoekstra, 2013). Investigations exploring suitability and effectiveness of ad personalization report that factors such as advertised product, data used for personalizing may influence the effectiveness of personalized ads (De Keyzer, Dens, & De Pelsmacker, 2015; Goldfarb & Tucker, 2011; van Doorn & Hoekstra, 2013). In this research we examine the impact of personalization triggers (PTs) on click through rate (CTR) of online banner ads across cultures. CTR data for 1345 unique ad copies (personalized) of an international hotel group screened in Japan and Middle East countries was used for this study. Data analyses revealed significant impact of PTs on CTRs. Analyses further revealed that – 1) usage of past purchase data impacts the CTR negatively, implying that customers respond negatively to ads showing hotel properties that they have previously visited/stayed in; 2) usage of search history data has a significant positive impact on CTR, suggesting that customers respond favourably towards ads showing hotel properties in destinations revealed from their search history. Interestingly, culture specific data such as local language elicit different responses in different cultures. While in Japan, language personalised ads i.e., ads in Japanese language fared poorly (negative impact on CTR) as compared to ads in English language (positive impact on CTR); in the Middle East it was the ads in English language that fared poorly (negative impact on CTR) compared to ads in Arabic. These results strongly suggest that the knowledge of PTs influence CTR and combining them with the right creative elements would help advertisers in improving customer engagement with ads, have a positive impact on CTR and even improve customer conversion. This would imply better returns for the resources spent on digital advertising. Findings from the study are true and reflective of the PTs (membership, brand affinity, destination, language) used in the ad campaign under study and limited to the cultures investigated. Future studies exploring other PTs in online hotel ads would help marketers in making a more informed decision while selecting data signals or PTs for personalizing digital banner ads for hotel brands.

In this research, we explore the role of interpersonal personalization in brand offerings in engendering the representations of ‘which dimension’ of brand personality (Aaker, 1997) and the subsequent effect on consumer attitude toward and connection with the brand (Fournier, 1998). Experiment 1 reveals that interpersonal personalization is different from customization with unique benefit in developing sincerity in brand personality. It demonstrates the positive effect of personalization on brand attitude through formulation of a sincere brand. Experiment 2 provides empirical evidence that personalization encouraging consumer’s involvement in the product design process allows consumer’s greater engagement with and investment towards the brand, which then manifests favorable brand attitude. The findings of this research offer insights on how brands can take advantage of personalization practice in their product or service offerings to engage consumers, and subsequently develop a close relationship with them. This research makes an important contribution to the brand personality literature through identifying a specific marketing approach (i.e., interpersonal personalization) as an effective strategy in developing a specific type of brand personality (i.e., sincerity). In sum, this research offers new view on how to manage close relationship with consumers, which is a significant practical implication for marketing managers.