An official website of the United States government
Social media embed rich but noisy signals of physical locations of their users. Accurately inferring a user's location can significantly improve the user's experience on the social media and enable the development of new location-based applications. This paper proposes a novel community-based approach for predicting the location of a user by using communities in the egonet of the user. We further propose both geographical proximity and structural proximity metrics to profile communities in the ego-net of a user, and then evaluate the effectiveness of each individual metric on real social media data. We discover that geographical proximity metrics, such as average/median haversine distance and community closeness, are strong indicators of a good community for geotagging. In addition, structural proximity metric conductance performs comparable to geographical proximity metrics while triangle participation ratio and internal density are weak location indicators. To the best of our knowledge, this is the first effort to infer the physical location of a user from the perspective of latent communities in the user's ego-net.
more »
« less
EXPANDING USER NEED FINDING THROUGH ABDUCTIVE REASONING
Prior research has shown the importance of latent user needs for enabling innovation in early product development phases. The success of a product is largely dependent on to what extent the product satisfies customer needs, and latent user needs play a significant role in impacting the way the product or service unexpectedly delights the user. Complications arise because traditional need finding methods are not able to account for the nuances of latent user needs. A user's need is multidimensional while traditional methods are built on deductive reasoning. The traditional method isolates parts of the user's needs, only pointing to what is deducible within its search space. To address this, we introduce abduction as a way to broaden traditional need finding methods. From a logic based argument it is shown that abduction accounts for the dimensionality of user needs by integrating various traditional need finding theories using design knowledge to isolate the latent need. This theoretical development shows that latent need finding must go beyond a deductive focus, to developing methods that are able to conjecture with the deduced facts in order to abduce the latent user need.
more »
« less
- Award ID(s):
- 2050052
- PAR ID:
- 10590701
- Publisher / Repository:
- Design Society
- Date Published:
- Journal Name:
- Proceedings of the Design Society
- Volume:
- 3
- ISSN:
- 2732-527X
- Page Range / eLocation ID:
- 1745 to 1754
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Reinforcement Learning from Human Feedback (RLHF) is a powerful paradigm for aligning foundation models to human values and preferences. However, current RLHF techniques cannot account for the naturally occurring differences in individual human preferences across a diverse population. When these differences arise, traditional RLHF frameworks simply average over them, leading to inaccurate rewards and poor performance for individual subgroups. To address the need for pluralistic alignment, we develop a class of multimodal RLHF methods. Our proposed techniques are based on a latent variable formulation - inferring a novel user-specific latent and learning reward models and policies conditioned on this latent without additional user-specific data. While conceptually simple, we show that in practice, this reward modeling requires careful algorithmic considerations around model architecture and reward scaling. To empirically validate our proposed technique, we first show that it can provide a way to combat under- specification in simulated control problems, inferring and optimizing user-specific reward functions. Next, we conduct experiments on pluralistic language datasets representing diverse user preferences and demonstrate improved reward function accuracy. We additionally show the benefits of this probabilistic framework in terms of measuring uncertainty, and actively learning user preferences. This work enables learning from diverse populations of users with divergent preferences, an important challenge that naturally occurs in problems from robot learning to foundation model alignment.more » « less
-
In a future of pervasive augmented reality (AR), AR systems will need to be able to efficiently draw or guide the attention of the user to visual points of interest in their physical-virtual environment. Since AR imagery is overlaid on top of the user's view of their physical environment, these attention guidance techniques must not only compete with other virtual imagery, but also with distracting or attention-grabbing features in the user's physical environment. Because of the wide range of physical-virtual environments that pervasive AR users will find themselves in, it is difficult to design visual cues that “pop out” to the user without performing a visual analysis of the user's environment, and changing the appearance of the cue to stand out from its surroundings. In this paper, we present an initial investigation into the potential uses of dichoptic visual cues for optical see-through AR displays, specifically cues that involve having a difference in hue, saturation, or value between the user's eyes. These types of cues have been shown to be preattentively processed by the user when presented on other stereoscopic displays, and may also be an effective method of drawing user attention on optical see-through AR displays. We present two user studies: one that evaluates the saliency of dichoptic visual cues on optical see-through displays, and one that evaluates their subjective qualities. Our results suggest that hue-based dichoptic cues or “Forbidden Colors” may be particularly effective for these purposes, achieving significantly lower error rates in a pop out task compared to value-based and saturation-based cues.more » « less
-
Throughout the mechanical design process, designers, the majority of whom are men, often fail to consider the needs of women, resulting in consequences ranging from inconvenience to increase risk of death or serious injury. Although these biases are well-studied in other fields of research, the mechanical design field lacks formal investigation into this phenomenon. In this study, undergraduate engineering students (n = 151) took a survey in which they read a Persona describing a student makerspace employee and a Walkthrough describing their interaction with the makerspace while completing a project. During the Walkthrough, the user encountered various obstacles, or Pain Points. Participants were asked to recall and evaluate the Pain Points that the user encountered, then evaluated their perceptions of the makerspace and user. Six different experimental conditions were used to investigate the impact of gender-stereotyped tasks and the gender of a user on designers’ interpretation of them and their needs. In addition to finding that the gender of a user impacted the way a task environment was perceived, results confirmed the presence of androcentrism, or “default man” assumptions, in the way designers view end users of unknown gender. Future work will explore methodologies of overcoming this bias so that designers are able to consider the diverse needs of a range of users.more » « less
-
Abstract Eliciting informative user opinions from online reviews is a key success factor for innovative product design and development. The unstructured, noisy, and verbose nature of user reviews, however, often complicate large-scale need finding in a format useful for designers without losing important information. Recent advances in abstractive text summarization has created the opportunity to systematically generate opinion summaries from online reviews to inform the early stages of product design and development. However, two knowledge gaps hinder the applicability of opinion summarization methods in practice. First, there is a lack of formal mechanisms to guide the generative process with respect to different categories of product attributes and user sentiments. Second, the annotated training datasets needed for supervised training of abstractive summarization models are often difficult and costly to create. This article addresses these gaps by (1) devising an efficient computational framework for abstractive opinion summarization guided by specific product attributes and sentiment polarities, and (2) automatically generating a synthetic training dataset that captures various degrees of granularity and polarity. A hierarchical multi-instance attribute-sentiment inference mode is developed for assembling a high-quality synthetic dataset, which is utilized to fine-tune a pretrained language model for abstractive summary generation. Numerical experiments conducted on a large dataset scraped from three major e-Commerce retail store for apparel and footwear products indicate the performance, feasibility, and potentials of the developed framework. Several directions are provided for future exploration in the area of automated opinion summarization for user-centered design.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Journal Article:
- https://doi.org/10.1017/pds.2023.175
