Social Situational Language Learning through an Online 3D Game
Learning Facilitaton
/
Culbertson, Gabriel
/
Wang, Shiyu
/
Jung, Malte
/
Andersen, Erik
Proceedings of the ACM CHI'16 Conference on Human Factors in Computing
Systems
2016-05-07
v.1
p.957-968
© Copyright 2016 ACM
Summary: Learning a second language is challenging. Becoming fluent requires learning
contextual information about how language should be used as well as word
meanings and grammar. The majority of existing language learning applications
provide only thin context around content. In this paper, we present work in
Crystallize, a language learning game that combines traditional learning
approaches with a situated learning paradigm by integrating a spaced-repetition
system within a language learning roleplaying game. To facilitate long-term
engagement with the game, we added a new quest paradigm, "jobs," that allow a
small amount of design effort to generate a large set of highly-scaffolded
tasks that grow iteratively. A large-scale evaluation of the language learning
game "in the wild" with a diverse set of 186 people revealed that the game was
not only engaging players for extended amounts of time but that players learned
an average of 8.7 words in an average of 40.5 minutes.
HandVis: Visualized Gesture Support for Remote Cross-Lingual Communication
Late-Breaking Works: Collaborative Technologies
/
Lin, Kuan-Yu
/
Yong, Seraphina
/
Wang, Shuo-Ping
/
Lai, Chien-Tung
/
Wang, Hao-Chuan
Extended Abstracts of the ACM CHI'16 Conference on Human Factors in
Computing Systems
2016-05-07
v.2
p.1236-1242
© Copyright 2016 ACM
Summary: Effective communication between those who are not fluent in a non-native
language can potentially be quite difficult. The common language selected to be
used throughout an exchange can encumber those who might not speak it as
proficiently as others. Remote communication further heightens the difficulty
since less channels are available for communication. We introduce HandVis, a
video conferencing interface that visualizes elements of hand gesture, such as
trajectory and amount. Gesture is intended to be a communicative tool that can
compensate for language deficits. The results of a user study indicate how
HandVis can be utilized constructively by less-proficient speakers during
cross-lingual communication.
Team Dating: A Self-Organized Team Formation Strategy for Collaborative
Crowdsourcing
Late-Breaking Works: Collaborative Technologies
/
Lykourentzou, Ioanna
/
Wang, Shannon
/
Kraut, Robert E.
/
Dow, Steven P.
Extended Abstracts of the ACM CHI'16 Conference on Human Factors in
Computing Systems
2016-05-07
v.2
p.1243-1249
© Copyright 2016 ACM
Summary: Online crowds have the potential to do more complex work in teams, rather
than as individuals. However, at such a large scale, team formation can be
difficult to coordinate. (How) can we rely on the crowd itself to organize into
effective teams? Our research explores a strategy for "team dating", a
self-organized crowd team formation approach where workers try out and rate
different candidate partners. In two online experiments, we find that team
dating affects the way that people select partners and how they evaluate them.
We use these results to draw useful conclusions for the future of team dating
and its implications for collaborative crowdsourcing.
Deep eye fixation map learning for calibration-free eye gaze tracking
New techniques and environments
/
Wang, Kang
/
Wang, Shen
/
Ji, Qiang
Proceedings of the 2016 Symposium on Eye Tracking Research &
Applications
2016-03-14
p.47-55
© Copyright 2016 ACM
Summary: The existing eye trackers typically require an explicit personal calibration
procedure to estimate subject-dependent eye parameters. Despite efforts in
simplifying the calibration process, such a calibration process remains
unnatural and bothersome, in particular for users of personal and mobile
devices. To alleviate this problem, we introduce a technique that can eliminate
explicit personal calibration. Based on combining a new calibration procedure
with the eye fixation prediction, the proposed method performs implicit
personal calibration without active participation or even knowledge of the
user. Specifically, different from traditional deterministic calibration
procedure that minimizes the differences between the predicted eye gazes and
the actual eye gazes, we introduce a stochastic calibration procedure that
minimizes the differences between the probability distribution of the predicted
eye gaze and the distribution of the actual eye gaze. Furthermore, instead of
using saliency map to approximate eye fixation distribution, we propose to use
a regression based deep convolutional neural network (RCNN) that specifically
learns image features to predict eye fixation. By combining the distribution
based calibration with the deep fixation prediction procedure, personal eye
parameters can be estimated without explicit user collaboration. We apply the
proposed method to both 2D regression-based and 3D model-based eye gaze
tracking methods. Experimental results show that the proposed method
outperforms other implicit calibration methods and achieve comparable results
to those that use traditional explicit calibration methods.
SINGA: Putting Deep Learning in the Hands of Multimedia Users
Best Paper Session
/
Wang, Wei
/
Chen, Gang
/
Dinh, Anh Tien Tuan
/
Gao, Jinyang
/
Ooi, Beng Chin
/
Tan, Kian-Lee
/
Wang, Sheng
Proceedings of the 2015 ACM International Conference on Multimedia
2015-10-26
p.25-34
© Copyright 2015 ACM
Summary: Recently, deep learning techniques have enjoyed success in various
multimedia applications, such as image classification and multi-modal data
analysis. Two key factors behind deep learning's remarkable achievement are the
immense computing power and the availability of massive training datasets,
which enable us to train large models to capture complex regularities of the
data. There are two challenges to overcome before deep learning can be widely
adopted in multimedia and other applications. One is usability, namely the
implementation of different models and training algorithms must be done by
non-experts without much effort. The other is scalability, that is the deep
learning system must be able to provision for a huge demand of computing
resources for training large models with massive datasets. To address these two
challenges, in this paper, we design a distributed deep learning platform
called SINGA which has an intuitive programming model and good scalability. Our
experience with developing and training deep learning models for real-life
multimedia applications in SINGA shows that the platform is both usable and
scalable.
SINGA: A Distributed Deep Learning Platform
Open Source Software Competition
/
Ooi, Beng Chin
/
Tan, Kian-Lee
/
Wang, Sheng
/
Wang, Wei
/
Cai, Qingchao
/
Chen, Gang
/
Gao, Jinyang
/
Luo, Zhaojing
/
Tung, Anthony K. H.
/
Wang, Yuan
/
Xie, Zhongle
/
Zhang, Meihui
/
Zheng, Kaiping
Proceedings of the 2015 ACM International Conference on Multimedia
2015-10-26
p.685-688
© Copyright 2015 ACM
Summary: Deep learning has shown outstanding performance in various machine learning
tasks. However, the deep complex model structure and massive training data make
it expensive to train. In this paper, we present a distributed deep learning
system, called SINGA, for training big models over large datasets. An intuitive
programming model based on the layer abstraction is provided, which supports a
variety of popular deep learning models. SINGA architecture supports both
synchronous and asynchronous training frameworks. Hybrid training frameworks
can also be customized to achieve good scalability. SINGA provides different
neural net partitioning schemes for training large models. SINGA is an Apache
Incubator project released under Apache License 2.
Hand-Object Sense: A Hand-held Object Recognition System Based on RGB-D
Information
Videos/Demos 1:
/
Lv, Xiong
/
Jiang, Shuqiang
/
Herranz, Luis
/
Wang, Shuang
Proceedings of the 2015 ACM International Conference on Multimedia
2015-10-26
p.765-766
© Copyright 2015 ACM
Summary: Hand-held objects play an important role in human-human and human-machine
interaction. It can be used as a reference for understanding user intentions or
user requirements. In this technical demonstration, we introduce an object
recognition system called Hand-Object Sense that can automatically recognize
the object held by user. This system first detects and segments the hand-held
object by exploiting skeleton information combined with depth information.
Second, in the object recognition stage, this system exploits features computed
in different ways and fuses them to improve the recognition accuracy. Our
system can recognize objects in real-time and have a good tolerance to angle
and scale transformation. Furthermore, it has a good generalization capability
for unknown objects.
Challenged Content Delivery Network: Eliminating the Digital Divide
Demos 2:
/
Hong, Hua-Jun
/
Wang, Shu-Ting
/
Tan, Chih-Pin
/
El-Ganainy, Tarek
/
Harras, Khaled
/
Hsu, Cheng-Hsin
/
Hefeeda, Mohamed
Proceedings of the 2015 ACM International Conference on Multimedia
2015-10-26
p.799-800
© Copyright 2015 ACM
Summary: We present a complete system, called Challenged Content Delivery Network
(CCDN), to efficiently deliver multimedia content to mobile users who live in
developing countries, rural areas, or over-populated cities with no or weak
network infrastructure. These mobile users do not have always-on Internet
access. We demo our CCDN, implemented on a Linux server, Raspberry Pi proxies,
and Android phones from three aspects: multimedia, networking, and machine
learning tools. We propose multiple optimization algorithm modules that compute
personalized distribution plans, and maximize the overall user experience. CCDN
allows people living in area with challenged networks access to multimedia
content, like news reports, using mobile devices, such as smartphones. This in
turn will help in eliminating the digital divide, which refers to information
inequality to persons with different Internet accessing abilities.
Toward Dual Roles of Users in Recommender Systems
Session 8D: Recommendation
/
Wang, Suhang
/
Tang, Jiliang
/
Liu, Huan
Proceedings of the 2015 ACM Conference on Information and Knowledge
Management
2015-10-19
p.1651-1660
© Copyright 2015 ACM
Summary: Users usually play dual roles in real-world recommender systems. One is as a
reviewer who writes reviews for items with rating scores, and the other is as a
rater who rates the helpfulness scores of reviews. Traditional recommender
systems mainly consider the reviewer role while not taking into account the
rater role. However, the rater role allows users to express their opinions
toward reviews about items; hence it may indirectly indicate their opinions
about items, which could be complementary to the reviewer role. Since most
real-world recommender systems provide convenient mechanisms for the rater
role, recent studies show that typically there are much more helpfulness
ratings from the rater role than item ratings from the reviewer role.
Therefore, incorporating the rater role of users may have the potentials to
mitigate the data sparsity and cold-start problems in traditional recommender
systems. In this paper, we investigate how to exploit dual roles of users in
recommender systems. In particular, we provide a principled way to exploit the
rater role mathematically and propose a novel recommender system DualRec, which
captures both the reviewer role and the rater role of users simultaneously for
recommendation. Experimental results on two real world datasets demonstrate the
effectiveness of the proposed framework, and further experiments are conducted
to understand the importance of the rater role of users in recommendation.
Experiences with eNav: a low-power vehicular navigation system
Low-power systems and devices
/
Hu, Shaohan
/
Su, Lu
/
Li, Shen
/
Wang, Shiguang
/
Pan, Chenji
/
Gu, Siyu
/
Al Amin, Md Tanvir
/
Liu, Hengchang
/
Nath, Suman
/
Choudhury, Romit Roy
/
Abdelzaher, Tarek F.
Proceedings of the 2015 International Conference on Ubiquitous Computing
2015-09-07
p.433-444
© Copyright 2015 ACM
Summary: This paper presents experiences with eNav, a smartphone-based vehicular GPS
navigation system that has an energy-saving location sensing mode capable of
drastically reducing navigation energy needs. Traditional navigation systems
sample the phone's GPS at a fixed rate (usually around 1Hz), regardless of
factors such as current vehicle speed and distance from the next navigation
waypoint. This practice results in a large energy consumption and unnecessarily
reduces the attainable length of a navigation session, if the phone is left
unplugged. The paper investigates two questions. First, would drivers be
willing to sacrifice some of the affordances of modern navigation systems in
order to prolong battery life? Second, how much energy could be saved using
straightforward alternative localization mechanisms, applied to complement GPS
for vehicular navigation? According to a survey we conducted of 500 drivers, as
much as 91% of drivers said they would like to have a vehicular navigation
application with an energy saving mode. To meet this need, eNav exploits
on-board accelerometers for approximate location sensing when the vehicle is
sufficiently far from the next navigation waypoint (or is stopped). A user
test-study of eNav shows that it results in roughly the same user experience as
standard GPS navigation systems, while reducing navigation energy consumption
by almost 80%. We conclude that drivers find an energy-saving mode on
phone-based vehicular navigation applications desirable, even at the expense of
some loss of functionality, and that significant savings can be achieved using
straightforward location sensing mechanisms that avoid frequent GPS sampling.
WISDOM: an efficient framework of predicting WLAN availability with cellular
fingerprints
Localization and navigation
/
Wang, Shuai
/
Yu, Xiaofeng
/
Xie, Junqing
Proceedings of the 2015 International Conference on Ubiquitous Computing
2015-09-07
p.951-962
© Copyright 2015 ACM
Summary: Mobile devices with both WLAN adapter and cellular capability, which are
also known as dual-mode mobile terminals, are facing various challenges and
problems in conventional WLAN discovery mechanisms, including inefficiency in
network discovery, unavoidable energy consumption for frequent WLAN scanning,
and privacy information leaking in network probing. In this paper, we propose a
novel framework called WISDOM (Wireless Indicator Supervised Data Offloading
Manipulation), which can efficiently predict the availability of appropriate
WLAN access points (APs) for mobile device without the need of turning on its
WLAN adapter in advance. WISDOM takes advantage of historical cellular
fingerprints (i.e., the pairs of Cell-ID and Received Signal Strength
Indicator) to directly model the WLAN coverage, and perform WLAN availability
prediction based on the models given a query cellular fingerprint. Similarity
and Classification methods are introduced to work in the framework as
prediction methods. We have developed a WISDOM prototype and performed
simulation and real field tests under various situations. The results showed
WISDOM along with the proposed predication methods could reach at least an
average of 80% in accuracy and saving 60% of power consumption on average for
mobile devices.
Listwise Collaborative Filtering
Session 4C: Classifying & Ranking
/
Huang, Shanshan
/
Wang, Shuaiqiang
/
Liu, Tie-Yan
/
Ma, Jun
/
Chen, Zhumin
/
Veijalainen, Jari
Proceedings of the 2015 Annual International ACM SIGIR Conference on
Research and Development in Information Retrieval
2015-08-09
p.343-352
© Copyright 2015 ACM
Summary: Recently, ranking-oriented collaborative filtering (CF) algorithms have
achieved great success in recommender systems. They obtained state-of-the-art
performances by estimating a preference ranking of items for each user rather
than estimating the absolute ratings on unrated items (as conventional
rating-oriented CF algorithms do). In this paper, we propose a new
ranking-oriented CF algorithm, called ListCF. Following the memory-based CF
framework, ListCF directly predicts a total order of items for each user based
on similar users' probability distributions over permutations of the items, and
thus differs from previous ranking-oriented memory-based CF algorithms that
focus on predicting the pairwise preferences between items. One important
advantage of ListCF lies in its ability of reducing the computational
complexity of the training and prediction procedures while achieving the same
or better ranking performances as compared to previous ranking-oriented
memory-based CF algorithms. Extensive experiments on three benchmark datasets
against several state-of-the-art baselines demonstrate the effectiveness of our
proposal.
Exploiting User and Business Attributes for Personalized Business
Recommendation
Short Papers
/
Lu, Kai
/
Zhang, Yi
/
Zhang, Lanbo
/
Wang, Shuxin
Proceedings of the 2015 Annual International ACM SIGIR Conference on
Research and Development in Information Retrieval
2015-08-09
p.891-894
© Copyright 2015 ACM
Summary: Data sparsity and cold-start are two major problems in personalized
recommendation. They are especially severe in business recommendation, because
business transactions are usually completed offline and customers generally do
not provide ratings after a transaction. Due to these two problems, matrix
factorization (MF) models, which are shown to be effective in many
recommendation tasks, are likely to fail on business recommendation tasks,
especially for new users and new items. In this paper, we propose an Integrated
Bias and Factorization Model (IBFM), which exploits user and business
attributes. The user attributes include demographic information, vote
information, point-of-interests; the business attributes include check-in
information, locations, business names, categories, etc. To handle the
cold-start problem, we employ a sampling strategy to generate the latent factor
vectors for new users and new businesses based on similar users/businesses. Our
methods are evaluated on the data set used in the RecSys 2013 Yelp business
rating prediction challenge. Experimental results show that our proposed
methods significantly outperform several existing state-of-the-art methods. In
particular, the single model IBFM performs the best in this challenge on both
public and private leaderboards.
An Innovation Design for Hazardous Chemical/Gases Disaster Detection and
Analysis Equipment by Using Cross-Cultural User Scenarios and Service Design
Cross-Cultural Design Methods and Case Studies
/
Wang, Sheng-Ming
/
Huang, Cheih Ju
/
Chou, Lun-Chang
/
Chen, Pei-Lin
CCD 2015: 7th International Conference on Cross-Cultural Design Methods,
Practice and Impact
2015-08-02
v.1
p.232-240
Keywords: Service design; Cross-Cultural scenarios; Usability; Hazardous
chemical/gases; Disaster management
© Copyright 2015 Springer International Publishing Switzerland
Summary: Unexpected releases of toxic, reactive, or flammable liquids and gases in
processes involving highly hazardous chemicals or gas explosions have been
reported for many years. The recent incident happened in Taiwan at 31st July,
2014 shows that a series of gas explosions occurred in the Cianjhen and Lingya
districts of Kaohsiung in Taiwan, following reports of gas leaks earlier that
night claimed 31 lives and injured other 309 people. In this study, we
organized an interdisciplinary team that contains scholars from university,
leaders from firefighter department, high rank officers from disaster
management agencies, researchers and project managers from research institute
and gases detector manufacture company and product designers to work together
to propose an innovation design for hazardous chemicals/gases detection and
analysis equipment. Based on the QFD analysis, operation for air detection is
the most important feature. The results shown in the QFD Matrix, was further
analyzed using a questionnaire that polled 6 inter-disciplinary experts in
order to collect the pair-wise comparison results in AHP. The top 3 feature
from the AHP are similar to the QFD weight: Air Type (20.05%), Air
Concentration (19.71%), and Air Detection (17.44%) The results of this research
point out that the innovation product design should also include the design of
service mechanism in order to meet users' requirement. For cross-cultural user
scenarios perspective, design thinking method that use diagram and pictures for
providing info-graphic results and the usability of user interface (UI) are two
major factors should be included in the design process. The conclusions of this
study suggest that the integration of product design and service design, and
the co-working mechanism among interdisciplinary team play very important role
in the innovation design for hazardous chemicals/gases detection and analysis
equipment.
Impact of Intermittent Stretching Exercise Animation on Prolonged-Sitting
Computer Users' Attention and Work Performance
Fitness and Well-Being Applications
/
Wang, Sy-Chyi
/
Chern, Jin-Yuan
HCI International 2015: 17th International Conference on HCI: Posters'
Extended Abstracts, Part II
2015-08-02
v.5
p.484-488
Keywords: Stretching exercise animation; Brainwave; Attention score; Work performance
© Copyright 2015 Springer International Publishing Switzerland
Summary: The prevailing use of computers and the Internet has contributed to popular
symptoms of visual impairment, musculoskeletal injuries, and even emotional
disorders nowadays. While certain ergonomics software packages have thus been
designed to avoid or relieve the symptoms, some studies raised concern about
possible decline in attention and work performance. This study aimed to explore
the effects of the computer stretch/massage software on extended computer
users' attention and work performance. The Neuroscience brainwave monitor was
used to evaluate the participants' attention. Thirty college students who work
more than 4 h a day in front of computer were recruited and evenly distributed
to two groups. The participants in the experimental group were asked to perform
the task on computer for 30 min with a stretch program on, which was set to
pop-up every 10 min for about 30 s each. The control group took no breaks or
interventions. The results show that the computer break software did not
decrease the participants' attention scores. Meanwhile the experimental group
demonstrated higher work performance scores. It is suggested that during
prolonged sitting computer work, breaks and body movements are necessary for
better attention and work performance.
New Research Methods for Media and Cognition Experiment Course
Designing the Playing Experience
/
Yang, Yi
/
Wang, Shengjin
/
Peng, Liangrui
DUXU 2015: Fourth International Conference on Design, User Experience, and
Usability, Part III: Interactive Experience Design
2015-08-02
v.3
p.327-334
Keywords: Media and cognition; Analysis of human brain; Human-computer interaction;
High-level talents; Investigation of project programming
© Copyright 2015 Springer International Publishing Switzerland
Summary: With the development of human-brain cognition and signal processing
techniques, there is more attention on media and cognitive disciplines,
especially focus on human-computer interaction and human's brain function
analysis. Electronic media is a new expression of human civilization, culture
and arts. Media and cognition experiment course is to complete the goal of
training talents through a large number of state-of-the-art methods. This paper
describes the understanding of the new practical engineering projects on media
and cognition course. Students were asked to complete several sets of practical
engineering courses. Some optional contents are also included. After this
training, we were able to select and train more high-level talents further. In
fact, this kind of practical engineering course can improve the students'
ability to grasp related knowledge points. Eventually they will have the
ability to plan projects and solve practical problems.
WikiMirs 3.0: A Hybrid MIR System Based on the Context, Structure and
Importance of Formulae in a Document
Session 7 -- Non-text Collections
/
Wang, Yuehan
/
Gao, Liangcai
/
Wang, Simeng
/
Tang, Zhi
/
Liu, Xiaozhong
/
Yuan, Ke
JCDL'15: Proceedings of the 2015 ACM/IEEE-CS Joint Conference on Digital
Libraries
2015-06-21
p.173-182
© Copyright 2015 ACM
Summary: Nowadays, mathematical information is increasingly available in websites and
repositories, such like ArXiv, Wikipedia and growing numbers of digital
libraries. Mathematical formulae are highly structured and usually presented in
layout presentations, such as PDF, LATEX and Presentation MathML. The
differences of presentation between text and formulae challenge traditional
text-based index and retrieval methods. To address the challenge, this paper
proposes an upgraded Mathematical Information Retrieval (MIR) system, namely
WikiMirs 3.0, based on the context, structure and importance of formulae in a
document. In WikiMirs 3.0, users can easily "cut" formulae and contexts from
PDF documents as well as type in queries. Furthermore, a novel hybrid indexing
and matching model is proposed to support both exact and fuzzy matching. In the
hybrid model, both context and structure information of formulae are taken into
consideration. In addition, the concept of formula importance within a document
is introduced into the model for more reasonable ranking. Experimental results,
compared with two classical MIR systems, demonstrate that the proposed system
along with the novel model provides higher accuracy and better ranking results
over Wikipedia.
Ariadne's Thread: Interactive Navigation in a World of Networked Information
WIP Theme: Search and Infoviz
/
Koopman, Rob
/
Wang, Shenghui
/
Scharnhorst, Andrea
/
Englebienne, Gwenn
Extended Abstracts of the ACM CHI'15 Conference on Human Factors in
Computing Systems
2015-04-18
v.2
p.1833-1838
© Copyright 2015 ACM
Summary: This work-in-progress paper introduces an interface for the interactive
visual exploration of the context of queries using the ArticleFirst database, a
product of OCLC. We describe a workflow which allows the user to browse live
entities associated with 65 million articles. In the on-line interface, each
query leads to a specific network representation of the most prevailing
entities: topics (words), authors, journals and Dewey decimal classes linked to
the set of terms in the query. This network represents the context of a query.
Each of the network nodes is clickable: by clicking through, a user traverses a
large space of articles along dimensions of authors, journals, Dewey classes
and words simultaneously. We present different use cases of such an interface.
This paper provides a link between the quest for maps of science and on-going
debates in HCI about the use of interactive information visualisation to
empower users in their search.
EmotiSphere: From Emotion to Music
Work-in-Progress: Poster Presentations
/
Chuang, Galen
/
Wang, Shelley
/
Burns, Sara
/
Shaer, Orit
Proceedings of the 2015 International Conference on Tangible and Embedded
Interaction
2015-01-15
p.599-602
© Copyright 2015 ACM
Summary: EmotiSphere is an interactive sensor-based musical instrument that generates
music based on a user's current emotional state. Interactions with EmotiSphere
draw upon everyday interactions with physical spherical objects, as well as on
familiar interactions with music players. EmotiSphere offers a novel way to
understand the relationship between emotion and music, and is aimed at people
who want to create music and express themselves but do not necessarily possess
skills in music composition. We describe the conceptualization and context of
EmotiSphere, as well as its technical implementation.
Attractive or Not?: Beauty Prediction with Attractiveness-Aware Encoders and
Robust Late Fusion
Posters 1
/
Wang, Shuyang
/
Shao, Ming
/
Fu, Yun
Proceedings of the 2014 ACM International Conference on Multimedia
2014-11-03
p.805-808
© Copyright 2014 ACM
Summary: Facial attractiveness is an ever-lasting issue in art and social science. It
also draws considerable attention from multimedia community recently. In this
paper, we develop a framework highlighting attractiveness-aware feature
extracted from a pair of auto-encoders to learn human-like assessment of facial
beauty. Our work is fully-automatic that does not require any landmark and puts
no restrictions on the faces' pose, expressions, and lighting conditions and
therefore is applicable on a larger and more diverse dataset. To this end,
first, a pair of auto-encoders is built respectively with beauty images and
non-beauty images, which can be used to extract attractiveness-aware features
by putting test images into both encoders. Second, we further enhance the
performance using an efficient robust low-rank fusion framework to integrate
the predicted confidence scores which are obtained based on certain kinds of
features. We show that our attractiveness-aware model with multiple layers of
auto-encoders produces appealing results and performs better than previous
appearance-based approaches.
A Cross-modal Multi-task Learning Framework for Image Annotation
KM Session 5: Classification II
/
Xie, Liang
/
Pan, Peng
/
Lu, Yansheng
/
Wang, Shixun
Proceedings of the 2014 ACM Conference on Information and Knowledge
Management
2014-11-03
p.431-440
© Copyright 2014 ACM
Summary: With the advance of internet, multi-modal data can be easily collected from
many social websites such as Wikipedia, Flickr, YouTube, etc. Images shared on
the web are usually associated with social tags or other textual information.
Although existing multi-modal methods can make use of associated text to
improve image annotation, the disadvantages of them are that associated text is
also required for a new image to be predicted. In this paper, we propose the
cross-modal multi-task learning (CMMTL) framework for image annotation. Labeled
and unlabeled multi-modal data are both levaraged for training in CMMTL, and it
finally obtains visual classifiers which can predict concepts for a single
image without any associated information. CMMTL integrates graph learning,
multi-task learning and cross-modal learning into a joint framework, where a
shared subspace is learned to preserve both cross-modal correlation and concept
correlation. The optimal solution of the proposed framework can be obtained by
solving a generalized eigenvalue problem. We conduct comprehensive experiments
on two real world image datasets: MIR Flickr and NUS-WIDE, to evaluate the
performance of the proposed framework. Experimental results demonstrate that
CMMTL obtains a significant improvement over several representative methods for
cross-modal image annotation.
Transfer Understanding from Head Queries to Tail Queries
KM Session 17: Web Data Mining
/
Song, Yangqiu
/
Wang, Haixun
/
Chen, Weizhu
/
Wang, Shusen
Proceedings of the 2014 ACM Conference on Information and Knowledge
Management
2014-11-03
p.1299-1308
© Copyright 2014 ACM
Summary: One of the biggest challenges of commercial search engines is how to handle
tail queries, or queries that occur very infrequently. Frequent queries, also
known as head queries, are easier to handle largely because their intents are
evidenced by abundant click-through data (query logs). Tail queries have little
historical data to rely on, which makes them difficult to be learned by ranking
algorithms. In this paper, we leverage knowledge from two resources to fill the
gap. The first is a general knowledgebase containing different granularities of
concepts automatically harnessed from the Web. The second is the click-through
data for head queries. From the click-through data, we obtain an understanding
of queries that trigger clicks. Then, we show that by extracting single or
multi-word expressions from both head and tail queries and mapping them to a
common concept space defined by the knowledgebase, we are able to transfer the
click information of the head queries to the tail queries. To validate our
approach, we conduct large scale experiments on two real data sets. One is a
mixture of head and tail queries, and the other contains pure tail queries. We
show that our approach effectively improves tail query search relevance.
Exploring Legal Patent Citations for Patent Valuation
KM Session 18: Data Mining Applications & Bioinformatics
/
Wang, Shuting
/
Lei, Zhen
/
Lee, Wang-Chien
Proceedings of the 2014 ACM Conference on Information and Knowledge
Management
2014-11-03
p.1379-1388
© Copyright 2014 ACM
Summary: Effective patent valuation is important for patent holders. Forward patent
citations, widely used in assessing patent value, have been considered as
reflecting knowledge flows, just like paper citations. However, patent
citations also carry legal implication, which is important for patent
valuation. We argue that patent citations can either be technological citations
that indicate knowledge transfer or be legal citations that delimit the legal
scope of citing patents. In this paper, we first develop citation-network based
methods to infer patent quality measures at either the legal or technological
dimension. Then we propose a probabilistic mixture approach to incorporate both
the legal and technological dimensions in patent citations, and an iterative
learning process that integrates a temporal decay function on legal citations,
a probabilistic citation network based algorithm and a prediction model for
patent valuation. We learn all the parameters together and use them for patent
valuation. We demonstrate the effectiveness of our approach by using patent
maintenance status as an indicator of patent value and discuss the insights we
learned from this study.
Exploit Latent Dirichlet Allocation for One-Class Collaborative Filtering
KM Track Posters
/
Zhang, Haijun
/
Li, Zhoujun
/
Chen, Yan
/
Zhang, Xiaoming
/
Wang, Senzhang
Proceedings of the 2014 ACM Conference on Information and Knowledge
Management
2014-11-03
p.1991-1994
© Copyright 2014 ACM
Summary: Previous work studied one-class collaborative filtering (OCCF) problems
including pointwise methods, pairwise methods, and content-based methods. The
fundamental assumptions made on these approaches are roughly the same. They
regard all missing values as negative. However, this is unreasonable since the
missing values actually are the mixture of negative and positive examples. A
user does not give a positive feedback on an item probably only because she/he
is unaware of the item, but in fact, she/he is fond of it. Furthermore,
content-based methods, e.g. collaborative topic regression (CTR), usually
require textual content information of items. This cannot be satisfied in some
cases. In this paper, we exploit latent Dirichlet allocation (LDA) model on
OCCF problem. It assumes missing values unknown and only models the observed
data, and it also does not need content information of items. In our model
items are regarded as words and users are considered as documents and the
user-item feedback matrix denotes the corpus. Experimental results show that
our proposed method outperforms the previous methods on various
ranking-oriented evaluation metrics.
INK: A Cloud-Based System for Efficient Top-k Interval Keyword Search
Demo Session 1
/
Li, Rui
/
Zhang, Xiao
/
Zhou, Xin
/
Wang, Shan
Proceedings of the 2014 ACM Conference on Information and Knowledge
Management
2014-11-03
p.2003-2005
© Copyright 2014 ACM
Summary: It is insufficient to search temporal text by only focusing on either time
attribute or keywords today as we pay close attention to the evolution of event
with time. Both temporal and textual constraints need to be considered in one
single query, called Top-k Interval Keyword Query (TIKQ). In this paper, we
presents a cloud-based system named INK that supports efficient execution of
TIKQs with appropriate effectiveness on Hadoop and HBase. In INK, an Adaptive
Index Selector (AIS) is devised to choose the better execution plan for various
TIKQs adaptively based on the proposed cost model, and leverage two novel
hybrid index modules (TriI and IS-Tree) to combine keyword and interval
filtration seamlessly.
