Nhật Long Lê - Academia.edu (original) (raw)

Papers by Nhật Long Lê

International journal of semantic computing, Dec 1, 2019

Customer reviews are an essential resource to reduce an online product’s uncertainty, which has b... more Customer reviews are an essential resource to reduce an online product’s uncertainty, which has been shown to be a critical factor for its purchase decision. Existing e-commerce platforms typically ask users to write free-form text reviews, which are sometimes augmented by a small set of predefined questions, e.g. “rate the product description’s accuracy from 1 to 5.” In this paper, we argue that this “passive” style of review solicitation is suboptimal in achieving low-uncertainty “review profiles” for products. Its key drawback is that some product aspects receive a very large number of reviews while other aspects do not have enough reviews to draw confident conclusions. Therefore, we hypothesize that we can achieve lower-uncertainty review profiles by carefully selecting which aspects users are asked to rate. To test this hypothesis, we propose various techniques to dynamically select which aspects to ask users to rate given the current review profile of a product. We use Bayesian inference principles to define reasonable review profile uncertainty measures; specifically, via an aspect’s rating variance. We compare our proposed aspect selection techniques to several baselines on several review profile uncertainty measures. Experimental results on two real-world datasets show that our methods lead to better review profile uncertainty compared to aspect selection baselines and traditional passive review solicitations. Moreover, we present and evaluate a hybrid solicitation method that combines the advantages of both active and passive review solicitations.

Information Systems, Nov 1, 2021

Document repositories often provide a keyword-based query interfaces to allow users to search for... more Document repositories often provide a keyword-based query interfaces to allow users to search for documents. These interfaces typically have rate limits or monetary cost per access operation. Constrained search interfaces include legal or medical data sources, social networks and the Web. We study the problem where a user has a set of input documents, and wants to discover other similar documents using a constrained search interface. Specifically, given a set of input documents and an access budget, we present principled techniques to generate a list of queries to submit. Our technique's key intuition is to compute the best set of queries to return the input documents, which, as we show experimentally, also return other relevant documents. We show that our techniques are superior to the state-of-the-art work, according to several intuitive document relevance metrics, on several real benchmark datasets. We show results for two problem variants: finding queries to return in the highest positions the input documents (Docs2Queries-Self problem) and other relevant documents (Docs2Queries-Sim problem).

Social Science Research Network, 2011

We develop an adaptive learning game to rethink efficient markets. We use the stochastically stab... more We develop an adaptive learning game to rethink efficient markets. We use the stochastically stable state of this game, which is a mixed Nash equilibrium, to form an adaptive expectation model that provides an estimate of the confidence interval for prices on the next day. The estimate is most accurate in the time of bubbles and crises, when rational expectations no longer fully hold.

RePEc: Research Papers in Economics, 2012

Social Science Research Network, 2013

We investigate a shift in market norm from an efficient state, where prices are unpredictable, to... more We investigate a shift in market norm from an efficient state, where prices are unpredictable, to the state where overconfidence spreads by contagion. Overconfidence means investors will buy more stocks that have enjoyed recent gains. It suggests that price shocks in a recent past may explain the signs of the price changes in the next period. To verify our conjecture, we run the Dickey – Fuller test for Dow Jones indexes and fail to reject the hypothesis of a unit root. But the residuals of the ARIMA process produced by this failure show some evidence that overconfidence, or more generally, overreaction, does play some role in determining price changes. Based on that evidence, we form a simple model to predict the direction of price changes in the next period. The forecasts fit the data reasonably well.

Searching through a large audio database for a specific sound can be a slow and tedious task with... more Searching through a large audio database for a specific sound can be a slow and tedious task with detrimental effects on creative workflow. Listening to each sample is time consuming, while textual descriptions or tags may be insufficient, unavailable or simply unable to meaningfully capturing certain sonic qualities. This paper explores the use of visual sketches that express the mental model associated with a sound to accelerate the search process. To achieve this, a study was conducted to collect data on how 30 people visually represent sound, by providing hand-sketched visual representations for a range of 30 different sounds. After augmenting the data to a sparse set of 855 samples, two different autoencoder were trained. The one finds similar sketches in latent space and delivers the associated audio files. The other one is a multimodal autoencoder combining both visual and sonic cues in a common feature space but lacks on having no audio input for the search task. These both were then used to implement and discuss a visual query-by-sketch search interface for sounds.

Social Science Research Network, 2015

Standard financial models assume that capital markets are fully efficient, which makes asset pric... more Standard financial models assume that capital markets are fully efficient, which makes asset prices unforecastable. In contrast, the behavioral finance argues that markets may not be efficient, at least in the short term, given the limits to arbitrage. Combining both strands of literature, our paper provides evidence to suggest that multiple states of market efficiency may exist. More precisely, in this multi-equilibria world, “the market” can transit from one state to another and a shift in market norm affects price movements in a near future. Our empirical analysis suggests a possibility of asset price predictability in the short term, based on the evolutionary market efficiency.

Social Science Research Network, 2014

This paper address the long-standing question of whether asset prices are predictable. The common... more This paper address the long-standing question of whether asset prices are predictable. The common view holds that daily prices fully incorporate all available information, and therefore price changes are unforecastable. This conclusion does not necessarily hold when the vast bulk of market trades are made by investors, who choose naïve price extrapolation rules over above fundamental-based rules. For that to happen, markets must suffer from two key problems. First, limits to arbitrage prevent informed speculators from setting prices right. Second, some rational speculators choose to bet not on fundamentals, but on future crowd behavior. And more important, their actions spread by contagion and become a norm. Under this norm, daily prices follow a random walk. But time series of price changes are co-integrated with time series of net demands generated by price extrapolation rules. As a result, prices are predictable in some degree.

Science & Technology Development Journal - Engineering and Technology

This is an experimental study to determine that the minimum fluidization velocity, the fluidized ... more This is an experimental study to determine that the minimum fluidization velocity, the fluidized bed height, and the pressure loss through the grain layer depend on the thickness of grain layers and moisture content of materials. The drying materials used in this study are fresh pepper and dried pepper of which moisture content is 54.2% and 12%, respectively. All the experiments are conducted with the thickness of the pepper layer ranging from 4mm to 44mm while the wind speed ranging from 5 m/s to 11m/s. From the result of experimental analysis, it can be seen that the minimum fluidization velocity of an air stream through fresh pepper ranges from 5.2m/s to 6,5m/s. The pressure drop of air stream through the fresh pepper is 225Pa while 147Pa for dry pepper, corresponding to the grain thickness of 37mm layer. By comparing the results determined from these experiments with computational models and some available correlations, the computational models of Fedorov, Baeyens-Geldart and Le...

IEEE Transactions on Robotics

There is a growing need for vertical takeoff and landing vehicles, including drones, which are sa... more There is a growing need for vertical takeoff and landing vehicles, including drones, which are safe to use and can adapt to collisions. The risks of damage by collision, to humans, obstacles in the environment, and drones themselves, are significant. This has prompted a search into nature for a highly resilient structure that can inform a design of propellers to reduce those risks and enhance safety. Inspired by the flexibility and resilience of dragonfly wings, we propose a novel design for a biomimetic drone propeller called Tombo propeller. Here, we report on the design and fabrication process of this biomimetic propeller that can accommodate collisions and recover quickly, while maintaining sufficient thrust force to hover and fly. We describe the development of an aerodynamic model and experiments conducted to investigate performance characteristics for various configurations of the propeller morphology and related properties, such as generated thrust force, thrust force deviation, collision force, recovery time, lift-to-drag ratio, and noise. Finally, we design and showcase a control strategy for a drone equipped with Tombo propellers that collides in midair with an obstacle and recovers from collision continuing flying. The results show that the maximum collision force generated by the proposed Tombo propeller is less than two-thirds that of a traditional rigid propeller, which suggests the concrete possibility to employ deformable propellers for drones flying in a cluttered environment. This research can contribute to the morphological design of flying vehicles for agile and resilient performance.

Asian Journal of International Law

Vietnam Journal of Endolaparoscopic Surgey

Tóm tắt Đặt vấn đề: Giãn đại tràng bẩm sinh là bệnh không có tế bào hạch thần kinh ở lớp cơ thành... more Tóm tắt Đặt vấn đề: Giãn đại tràng bẩm sinh là bệnh không có tế bào hạch thần kinh ở lớp cơ thành đại tràng, bệnh thường gọi là megacolon hay Hichsprung. Người bệnh chủ yếu được phát hiện bệnh và can thiệp phẫu thuật khi còn nhỏ tuổi. Một số trường hợp, (thường do đoạn vô hạch ngắn, ở phần thấp trực tràng) nên các triệu chứng không điển hình, diễn biến bệnh kéo dài, đại trực tràng giãn nhiều, khối phân to, rắn, khó khăn cho việc điều trị và phẫu thuật. Đối tượng và phương pháp nghiên cứu: Tất cả người bệnh trên 16 tuổi chẩn đoán xác định là giãn đại trực tràng điều trị bằng phẫu thuật trong giai đoạn 1/2015 đến 12/2020 tại Bệnh viện Hữu nghị Việt Đức. Nghiên cứu hồi cứu mô tả cắt ngang. Kết quả: 41 người bệnh được phẫu thuật, tuổi trung bình 27,7 ± 11,3. 19 người bệnh nam (46,3%), 22 người bệnh nữ (53,7%). Mổ nội soi 10 người bệnh (24,4%), mổ mở 31 người bệnh (75,6%). Mổ 1 thì 16 người bệnh (39%), mổ 2 thì 3 người bệnh, mổ 3 thì 18 người bệnh (43,9%), 4 người bệnh phẫu thuật Hartman...

2021 IEEE 15th International Conference on Semantic Computing (ICSC)

Most existing commercial goal-oriented chatbots are diagram-based; i.e., they follow a rigid dial... more Most existing commercial goal-oriented chatbots are diagram-based; i.e., they follow a rigid dialog flow to fill the slot values needed to achieve a user's goal. Diagram-based chatbots are predictable, thus their adoption in commercial settings; however, their lack of flexibility may cause many users to leave the conversation before achieving their goal. On the other hand, state-of-the-art research chatbots use Reinforcement Learning (RL) to generate flexible dialog policies. However, such chatbots can be unpredictable, may violate the intended business constraints, and require large training datasets to produce a mature policy. We propose a framework that achieves a middle ground between the diagram-based and RL-based chatbots: we constrain the space of possible chatbot responses using a novel structure, the chatbot dependency graph, and use RL to dynamically select the best valid responses. Dependency graphs are directed graphs that conveniently express a chatbot's logic by defining the dependencies among slots: all valid dialog flows are encapsulated in one dependency graph. Our experiments in several domains show that our framework quickly adapts to user characteristics and achieves up to 23.77% improved success rate compared to a state-of-the-art RL model.

2019 IEEE International Conference on Humanized Computing and Communication (HCC), 2019

Transactional chatbots have become popular today, as they can automate repetitive transactions su... more Transactional chatbots have become popular today, as they can automate repetitive transactions such as making an appointment or buying a ticket. As users interact with a chatbot, rich chat logs are generated to evaluate and improve the effectiveness of the chatbot, which is the ratio of chats that lead to a successful state such as buying a ticket. A fundamental operation to achieve such analyses is the clustering of the chats in the chat log, which requires effective distance functions between a pair of chat sessions. In this paper, we propose and compare various distance measures for individual messages as well as whole sessions. We evaluate these measures using user studies on Mechanical Turk, where we first ask users to use our chatbots, and then ask them to judge the similarity of messages and sessions. Finally, we provide anecdotal results showing that our distance functions are effective in clustering messages and sessions.

2019 IEEE 13th International Conference on Semantic Computing (ICSC), 2019

Customer reviews have become an essential resource when people search for goods or services on th... more Customer reviews have become an essential resource when people search for goods or services on the Internet. Previous work has shown that reducing a product's uncertainty is critical to its purchase decision. Thus, reviews are more effective when they reduce a product's uncertainty. Existing e-commerce platforms typically ask users to write free-form text reviews, which are sometimes augmented by a small set of predefined questions, e.g., "rate the product description's accuracy from 1 to 5." In this paper, we argue that this "passive" style of review solicitation is suboptimal in achieving low-uncertainty "review profiles" for products. Its key drawback is that some product aspects receive a very large number of reviews while other aspects do not have enough reviews to draw confident conclusions. Therefore, we hypothesize that we can achieve lower-uncertainty review profiles by carefully selecting which aspects users are asked to rate. To test this hypothesis, we propose various techniques to dynamically select which aspects to ask users to rate given the current review profile of a product. We use Bayesian principles to define reasonable review profile uncertainty measures; specifically, we apply Bayesian inference to measure an aspect's rating variance. We compare our proposed aspect selection techniques to several baselines on several review profile uncertainty measures. Experimental results on two real-world datasets show that our methods lead to better review profile uncertainty compared to aspect selection baselines and traditional passive review solicitations.

International Journal of Semantic Computing, 2021

Most existing commercial goal-oriented chatbots are diagram-based; i.e. they follow a rigid dialo... more Most existing commercial goal-oriented chatbots are diagram-based; i.e. they follow a rigid dialog flow to fill the slot values needed to achieve a user’s goal. Diagram-based chatbots are predictable, thus their adoption in commercial settings; however, their lack of flexibility may cause many users to leave the conversation before achieving their goal. On the other hand, state-of-the-art research chatbots use Reinforcement Learning (RL) to generate flexible dialog policies. However, such chatbots can be unpredictable, may violate the intended business constraints, and require large training datasets to produce a mature policy. We propose a framework that achieves a middle ground between the diagram-based and RL-based chatbots: we constrain the space of possible chatbot responses using a novel structure, the chatbot dependency graph, and use RL to dynamically select the best valid responses. Dependency graphs are directed graphs that conveniently express a chatbot’s logic by definin...

Biomedical Optics Express, 2020

The structure of brain regions is assumed to correlate with their function, but there are very fe... more The structure of brain regions is assumed to correlate with their function, but there are very few instances in which the relationship has been demonstrated in the live brain. This is due to the difficulty of simultaneously measuring functional and structural properties of brain areas, particularly at cellular resolution. Here, we performed label-free, third-harmonic generation (THG) microscopy to obtain a key structural signature of cortical areas, their effective attenuation lengths (EAL), in the vertical columns of functionally defined primary visual cortex and five adjacent visual areas in awake mice. EALs measured by THG microscopy in the cortex and white matter showed remarkable correspondence with the functional retinotopic sign map of each area. Structural features such as cytoarchitecture, myeloarchitecture and blood vessel architecture were correlated with areal EAL values, suggesting that EAL is a function of these structural features as an optical property of these areas...

Applied Sciences, 2020

A person’s gait is a behavioral trait that is uniquely associated with each individual and can be... more A person’s gait is a behavioral trait that is uniquely associated with each individual and can be used to recognize the person. As information about the human gait can be captured by wearable devices, a few studies have led to the proposal of methods to process gait information for identification purposes. Despite recent advances in gait recognition, an open set gait recognition problem presents challenges to current approaches. To address the open set gait recognition problem, a system should be able to deal with unseen subjects who have not included in the training dataset. In this paper, we propose a system that learns a mapping from a multimodal time series collected using insole to a latent (embedding vector) space to address the open set gait recognition problem. The distance between two embedding vectors in the latent space corresponds to the similarity between two multimodal time series. Using the characteristics of the human gait pattern, multimodal time series are sliced i...