Twitter sentiment analysis with a deep neural network: An enhanced approach using user behavioral information

Sentiment analysis on social media such as Twitter has become a very important and challenging task. Due to the characteristics of such data—tweet length, spelling errors, abbreviations, and special characters—the sentiment analysis task in such an environment requires a non-traditional approach. Moreover, social media sentiment analysis is a fundamental problem with many interesting applications. Most current social media sentiment classification methods judge the sentiment polarity primarily according to textual content and neglect other information on these platforms. In this paper, we propose a neural network model that also incorporates user behavioral information within a given document (tweet). The neural network used in this paper is a Convolutional Neural Network (CNN). The system is evaluated on two datasets provided by the SemEval-2016 Workshop. The proposed model outperforms current baseline models (including Naive Bayes and Support Vector Machines), which shows that going beyond the content of a document (tweet) is beneficial in sentiment classification, because it provides the classifier with a deep understanding of the task.     

Object affordance based multimodal fusion for natural Human-Robot interaction

Spoken language based natural Human-Robot Interaction (HRI) requires robots to have the ability to understand spoken language, and extract the intention-related information from the working scenario. For grounding the intention-related object in the working environment, object affordance recognition could be a feasible way. To this end, we propose a dataset and a deep CNN based architecture to learn the human-centered object affordance. Furthermore, we present an affordance based multimodal fusion framework to realize intended object grasping according to the spoken instructions of human users. The proposed framework contains an intention semantics extraction module which is employed to extract the intention from spoken language, a deep Convolutional Neural Network (CNN) based object affordance recognition module which is applied to recognize human-centered object affordance, and a multimodal fusion module which is adopted to bridge the extracted intentions and the recognized object affordances. We also complete multiple intended object grasping experiments on a PR2 platform to validate the feasibility and practicability of the presented HRI framework.     

真实场景搜索中的眼动返回抑制

眼动返回抑制是指眼睛返回至先前注视过的位置(客体)时眼跳潜伏期长、可能性低的一种抑制现象。近年来, 研究者对真实场景搜索中的眼动返回抑制进行了深入的研究。本文介绍了相关研究的经典实验范式、主要发现及其理论解释, 探讨了眼动返回抑制的容量、眼动返回抑制是否具有任务特异性、场景搜索中是否存在眼动返回促进以及眼动返回抑制的神经基础等问题, 并指出未来该领域研究的方向和途径。     

局部亮度与语义关系对真实情景识别的影响

以真实情景图片为材料,探讨局部亮度与语义关系对情景识别的影响,以及情景背景线索和语义关系的获得。实验一结果显示,被试对有无语义关系情景图片的识别反应时存在显著差异,且对拼接图片的识别正确率更高,反应时更短。实验二探讨了局部亮度与呈现时间变化条件下情景背景线索和语义关系的获得。结果发现,情景语义关系和背景信息的获得受到局部亮度变化和呈现时间的影响。这表明,语义关系和呈现时间对情景识别存在显著影响,而且这种影响与情景背景线索和语义关系的获得有关。呈现时间越长,越易获得情景语义关系。局部亮度变化越小,背景线索越易获得,呈现时间的影响越小。     

To infinity and beyond: Children generalize the successor function to all possible numbers years after learning to count

Recent accounts of number word learning posit that when children learn to accurately count sets (i.e., become “cardinal principle” or “CP” knowers), they have a conceptual insight about how the count list implements the successor function – i.e., that every natural number n has a successor defined as n + 1 (Carey, 2004, 2009; Sarnecka & Carey, 2008). However, recent studies suggest that knowledge of the successor function emerges sometime after children learn to accurately count, though it remains unknown when this occurs, and what causes this developmental transition. We tested knowledge of the successor function in 100 children aged 4 through 7 and asked how age and counting ability are related to: (1) children’s ability to infer the successors of all numbers in their count list and (2) knowledge that all numbers have a successor. We found that children do not acquire these two facets of the successor function until they are about 5½ or 6 years of age – roughly 2 years after they learn to accurately count sets and become CP-knowers. These findings show that acquisition of the successor function is highly protracted, providing the strongest evidence yet that it cannot drive the cardinal principle induction. We suggest that counting experience, as well as knowledge of recursive counting structures, may instead drive the learning of the successor function.     

Characterizing Human Expertise Using Computational Metrics of Feature Diagnosticity in a Pattern Matching Task

Forensic evidence often involves an evaluation of whether two impressions were made by the same source, such as whether a fingerprint from a crime scene has detail in agreement with an impression taken from a suspect. Human experts currently outperform computer‐based comparison systems, but the strength of the evidence exemplified by the observed detail in agreement must be evaluated against the possibility that some other individual may have created the crime scene impression. Therefore, the strongest evidence comes from features in agreement that are also not shared with other impressions from other individuals. We characterize the nature of human expertise by applying two extant metrics to the images used in a fingerprint recognition task and use eye gaze data from experts to both tune and validate the models. The Attention via Information Maximization (AIM) model (Bruce & Tsotsos, 2009) quantifies the rarity of regions in the fingerprints to determine diagnosticity for purposes of excluding alternative sources. The CoVar model (Karklin & Lewicki, 2009) captures relationships between low‐level features, mimicking properties of the early visual system. Both models produced classification and generalization performance in the 75%–80% range when classifying where experts tend to look. A validation study using regions identified by the AIM model as diagnostic demonstrates that human experts perform better when given regions of high diagnosticity. The computational nature of the metrics may help guard against wrongful convictions, as well as provide a quantitative measure of the strength of evidence in casework.     

上下文预期在快速场景识别中的作用

本研究首先考察上下文预期是否能够影响快速场景识别,进而通过比较不同空间频率的场景信息来探究上下文预期在快速场景识别不同阶段的作用。共包括3个实验:实验1采用双眼竞争范式,考察对快速呈现场景刺激的主观选择是否受到上下文预期的影响;实验2采用词汇分类与快速场景识别的双任务范式,比较预期与非预期条件下快速场景识别绩效的差异;实验3分别以低频信息(实验3a)和高频信息(实验3b)作为实验材料,探讨预期效应的作用阶段。结果发现:上下文预期可以影响快速场景识别过程中的主观选择与反应绩效;对不同空间频率的场景信息,场景刺激与预期相一致时均可以提高辨别力,但预期对反应偏好的影响只发生在对高空间频率场景信息的加工阶段。因此,上下文预期在快速场景识别不同加工阶段的作用不同,快速场景识别需要结合不同空间频率信息加工的结果。     

Do children learn the integers by induction?

According to one theory about how children learn the meaning of the words for the positive integers, they first learn that "one," "two," and "three" stand for appropriately sized sets. They then conclude by inductive inference that the next numeral in the count sequence denotes the size of sets containing one more object than the size denoted by the preceding numeral. We have previously argued, however, that the conclusion of this Induction does not distinguish the standard meaning of the integers from nonstandard meanings in which, for example, "ten" could mean set sizes of 10, 20, 30,... elements. Margolis and Laurence [Margolis, E., & Laurence, S. (2008). How to learn the natural numbers: Inductive inference and the acquisition of number concepts. Cognition, 106, 924-939] believe that our argument depends on attributing to children "radically indeterminate" concepts. We show, first, that our conclusion is compatible with perfectly determinate meanings for "one" through "three." Second, although the inductive inference is indeed indeterminate - which is why it is consistent with nonstandard meanings - making it determinate presupposes the constraints that the inference is supposed to produce.     

Revising the concept of lawhood: special sciences and natural kinds

The Kripkean conception of natural kinds (kinds are defined by essences that are intrinsic to their members and that lie at the microphysical level) indirectly finds support in a certain conception of a law of nature, according to which generalizations must have unlimited scope and be exceptionless to count as laws of nature. On my view, the kinds that constitute the subject matter of special sciences such as biology may very well turn out to be natural despite the fact that their essences fail to be microphysical or micro-based. On the causal conception of natural kinds I privilege, the naturalness of a kind is a function of the fact that it figures prominently in at least one causal law. However, there is a strong tendency prevailing among contemporary philosophers to assume that, in order to count as proper laws generalizations must be expectionless. Since most generalizations tracked down by the special sciences turn out not to fulfill these criteria, what this conception of a law implies is that most of the generalizations the special sciences trade in are not proper laws. It follows that, on this view, most if not all of the kinds the special sciences dealing with turn out not to constitute natural kinds, understood as kinds to which bona fide laws apply. In order to establish that the non-microstructurally defined kinds that fall within the domain of enquiry of the special sciences are eligible for the status of natural kind, I must therefore establish that generalizations needn’t have unlimited scope and be exceptionless to count as laws of nature. This is precisely what I seek to do in this paper. I begin by arguing that the question “what is a law of nature?” is most naturally interpreted as the question “what features must generalizations exhibit in order to ground scientific explanations?” and by offering reasons to believe that generalizations needn’t be exceptionless and have unlimited scope to play the crucial role laws have been thought to play in scientific explanation. Drawing on Sandra Mitchell [Mitchell, S. (2000). Philosophy of Science, 67, 242–265] and James Woodward’s [Woodward, J. (1997). Philosophy of science, 64 (proceedings), 524–541; Woodward, J. (2000). British Journal for the philosophy of science, 51(2), 197–254; Woodward, J. (2001). Philosophy of science, 68, 1–20] work, I subsequently develop an alternative account of the criteria generalizations must satisfy in order to count as laws of nature, which at least some of the generalizations of the special sciences turn out to fulfill. I thus give credence to the idea that at least some of the kinds that fall within the domain of the special sciences figure in laws of nature, and I thereby restore the possibility that some special science kinds deserve to be deemed natural.     

“Eat your Hamburger!”—“No, I don’t Want to!” Argumentation and Argumentative Development in the Context of Dinner Conversation in Twenty Swedish Families

The aim of the present study was to analyse family dinners as context of argumentation and argumentative development by using a context-sensitive model of basic argumentative structures in every day conversations. The data consisted of 40 argumentative sequences in dinner conversations in twenty Swedish families with children aged 7 to 17 years. The families were divided in two groups depending on the children's ages (10–11 years with younger siblings and 10–12 years with older siblings). The model revealed characteristic structures of argumentation appearing as co-text and suggested differences between family groups depending on contextual factors such as age of the children. The groups of older children produced longer argumentative sequences, more exchanges per sequence and higher rate of turns. The older children also engaged in non-instrumental deliberations and disputations significantly more often and they performed more elaborated expansions (through a higher quantity of backing arguments). The groups of younger children on the other hand were more often involved in negotiations on topics relevant in the immediate context. Less expected was, however, the lack of more complex and varied arguments, even in the groups of older children.