Article

Natural Language Processing based Context Sensitive, Content Specific Architecture & its Speech based Implementation for Smart Home Applications

Authors:
  • Director, IIIT Kottayam, Kerala, India Institute of National Importance
To read the full-text of this research, you can request a copy directly from the authors.

Abstract

The one of the upcoming research stream of Computer Science and Engineering is Natural Language Processing (NLP), which is widely being used in design of Interfaces for Human-Computer Interaction. One of the basic applications of NLP, is design of "Smart Homes", in which; based on user input certain actions can be performed either locally inside the house or globally outside the house. The smart homes designed are either based on remote input provided to the system from one place in the house or limited to only certain type of actions which the system can handle. The current state of art of the Smart Home design methodologies, does not includes the design of customized systems capable of handling inputs from different gender i.e., in different pitches, similarly the methodologies does not provides facilities to handle input in different language. The systems existing are not capable of understanding the context of situation and determining actions based on context. The paper describes a Smart Home application which can be used by elderly people living alone in the home to serve their basic needs, which specifically includes security issues. The system is initially based on generic architecture, and can be further customized to user needs. The main component of the generic architecture is ability to fix a certain language or set of languages in which the input will be provided to the system. This enables the user to interact with the system in multiple languages, thus the specific instruction set is not limited to one language. The interface designed is based on speech input, can handle multiple languages and is not gender specific. The system designed is also "Context Specific", to understand the context of current state in which input is given to the system and perform the necessary action. The context understanding feature makes the system more specific to understand the urgency of action to be performed based on input. The system can deployed for a building incorporating a wireless sensor network, and provide a high quality, efficient context-sensitive data transmission facility.

No full-text available

Request Full-text Paper PDF

To read the full-text of this research,
you can request a copy directly from the authors.

... Moreover, since the goal samples are stored as text, a wider number of goal samples can be stored and identified. Chandak et al. [5] propose a smart home system to meet the basic requirements of elderly people using natural language processing (NLP) techniques. The system uses a list of keywords and identifies their synonyms using Wordnet. ...
... However, this approach has limitations in terms of understanding the indirect intent in users' goal, recognizing utterances with complex semantic structure e.g., negative goals that invert the meaning of the utterance or terms that have overlap within different contexts. Mofrad et al. [16] present a similar approach to [5], with the difference that it extracts two types of entities in an utterance, namely action and device. The prototype presented only supports one device and four actions: "play", "pause", "resume" and "stop". ...
Conference Paper
Full-text available
One of the main challenges of the Internet of Things (IoT) is to enable end-users without technical experience to use, control or monitor smart devices. However, enabling end-users to interact with these smart devices in an intuitive and natural way becomes increasingly important as they become more pervasive in our homes, workplaces and public environments. Voice-based interfaces are the emerging trend to provide a more natural human-device interaction in smart environments. Such interfaces require Natural Language Understanding (NLU) approaches to identify the meaning of end-users' voice inputs. Designing voice interfaces that are not limited to a small, fixed set of pre-defined commands is far from trivial. Existing voice-based solutions in the smart home domain either restrict the end-users to follow a strict language pattern, do not support indirect goals, require a large training dataset, or need a voice assistant located in the cloud. In this paper, we propose an approach for understanding end-users goals from voice inputs in smart homes. Our approach alleviates the need for end-users to learn or remember concrete operations of the devices and specific words/pattern structures rather it enables them to control their smart homes based on the desired goals (effects). We evaluate the approach through application to a collection of 253 goals from real end-users and report on quality metrics. The results demonstrate that our solution provides a good accuracy, high precision and acceptable recall for understanding end-users goals in the smart home domain.
... These pairs are known as edges, arcs, or lines for an undirected graph and as arrows, directed edges, directed arcs, or directed lines for a directed graph. The vertices may be part of the graph structure, or may be external entities represented by integer indices or reference [3,4]. A graph data structure may also associate to each edge some edge value, such as a symbolic label or a numeric attribute (cost, capacity, length, etc.).Many operations can be performed on graphs like adding an edge, removing a vertex, checking if 2 vertices are adjacent, etc. ...
... At each intersection there is a node. In such a model there are natural tasks such as searching for the shortest path between two intersections, checking whether there is a road between two intersections, checking for a loop (if we can turn and go back to the starting position) searching for a path with a minimum number of turns, etc. [3,4] -Computer network can be modelled by an undirected graph, whose vertices correspond to the computers in the network, and the edges correspond to the communication channels between the computers. To the edges different numbers can be compared, such as channel capacity or speed of the exchange, etc. ...
Article
Full-text available
A Graph has two main components: Vertices and Edges. The vertices are connected using edges. There are two types of graphs, directed and undirected. The major application of graph is representing network on paper. The cost involved in converting paper-based network to actual cable-based network is majorly controlled by cables required for connection. The cost can be reduced if the length of cable can be reduced. The paper describes the methodology to compute Steiner point. Using Steiner point, it is possible to modify the position of vertices, so as to reduce the cable length, keeping the vertex connectivity intact. The paper describes the implementation of Steiner point on the graph, with number of vertices as 3, 4, 5 and 6. The presented work can be extended for graph with any number of vertices. It is an optimization approach to reduce the cable size and cost of network implementation.
... Then, the registered information appliances will execute various commands that are received from users via social networking services. To realize such smart homes, information appliances should understand users' natural language commands which are in the form of short text messages (e.g., tweets and recognized speech inputs) [2]. Natural language processing (NLP) techniques can be used to convert statements of a natural language into the statements of a highly reduced topic-oriented language, a formal language that information appliances can understand [3]. ...
... In spite of these hardware limitations, lightweight NLP applications (e.g., a tweet clustering system [4], an appointment extraction system [5], a news summarization system (http://summly.com), and a translation assistance system [6]) and NLP-based smart home applications (e.g., a smart home security system [2] and a smart home social networking system [1]) have been successfully developed. To implement these NLP-based applications, words in a sentence should first be tokenized. ...
Article
Full-text available
In smart homes, information appliances interact with residents via social network services. To capture residents’ intentions, the information appliances should analyze short text messages entered typically through small mobile devices. However, most information appliances have hardware constraints such as small memory, limited battery capacity, and restricted processing power. Therefore, it is not easy to embed intelligent applications based on natural language processing (NLP) techniques, which traditionally require large memory and high-end processing power, into information appliances. To overcome this problem, lightweight NLP modules should be implemented. We propose an automatic word spacing system, the first step module of NLP for many languages with their own word spacing rules, which is designed for information appliances with limited hardware resources. The proposed system consists of a word spacing dictionary and a pattern-matching module. When a sentence is entered, the pattern-matching module inserts spaces by simply looking up the word spacing dictionary in a back-off manner. In comparative experiments with previous models, the proposed method showed low memory usage (0.79 MB) and high character-unit accuracy (0.9460) without requiring complex arithmetical computations. On the basis of these experiments, we conclude that the proposed system is suitable for information appliances with many hardware limitations.
... In 2010, Chandak and Dharaskar reported an attempt to implement speech-based controls for a context-sensitive, content-specific Smart Home architecture based on natural language processing [38]. The key to their system was the ability of the user to customize the specific language or languages to be used for input. ...
... New devices cannot be added without customization [52,104]. Furthermore, a different sort of customization allows a personal experience for the user [32,38], but not for more than one user [33], and not in a manner that is easily adaptable [40,57,58,127]. Generally, the systems are standardized [7] and require that the user(s) adapt themselves to their environment [9,54], in direct opposition to Weiser's intent [25,26]. ...
Thesis
Full-text available
Rather than adapt human behaviour to suit a life surrounded by computerized systems, is it possible to adapt the systems to suit humans? Mark Weiser called for this fundamental change to the design and engineering of computer systems nearly twenty years ago. We believe it is possible and offer a series of related theoretical developments and practical experiments designed in an attempt to build a system that can meet his challenge without resorting to black box design principles or Wizard of Oz protocols. This culminated in a trial involving 32 participants, each of whom used two different multi-modal interactive techniques, based on our novel interaction paradigm, to intuitively control nine distributed devices in a smart home setting. The theoretical work and practical developments have led to our proposal of seven contributions to the state of the art.
... Smart Cities are rapidly evolving with the integration of Internet of Things (IoT) devices and technologies [1][2][3][4]. Data-driven strategies are used by these urban centers to improve efficiency, sustainability, and people' quality of life [5][6][7][8]. However, security and anomaly detection are becoming increasingly difficult for Smart Cities as they become more linked. ...
Article
Full-text available
With the increasing deployment of Internet of Things (IoT) devices in Smart Cities, ensuring the security and integrity of these systems is of paramount importance. This study focuses on the detection of anomalies and prevention of security attacks in Smart City environments, specifically in the context of traffic management. We compare the performance of three machine learning algorithms, namely logistic boosted algorithms, random forest, and support vector machines, using a dataset collected from various Smart City devices. Through comprehensive evaluation metrics such as accuracy, precision, recall, and F1 score, we analyze the efficacy of each algorithm. Our results indicate that the logistic boosted algorithm outperforms the other approaches in accurately identifying data outliers and achieving high performance across all evaluation metrics. Notably, it excels in handling imbalanced datasets and leveraging ensemble methods to enhance overall performance. While the random forest algorithm demonstrates exceptional performance in all metrics, the support vector machines approach exhibits superior recall but comparatively lower accuracy, precision, and F1 score. Moreover, the logistic boosted method showcases fewer misclassifications in the confusion matrices, making it highly effective for anomaly detection. The area under the ROC curve is highest for the logistic boosted algorithm (0.98), further emphasizing its effectiveness in filtering out anomalies. The random forest algorithm achieves the highest area under the curve (0.982), while the support vector machines method shows the lowest (0.968). Based on our findings, we conclude that the logistic boosted algorithm is the most effective for monitoring Smart City systems, followed by the random forest approach. However, the support vector machines method can be a viable alternative if recall is the primary performance metric of concern. Future research can focus on comparing the efficacy of different machine learning approaches for anomaly detection in Smart City environments and exploring the impact of unique dataset properties on algorithm performance. Additionally, exploring the potential of deep learning algorithms for traffic anomaly detection in Smart City systems could be a promising area of study. This study contributes to the advancement of state-of-the-art anomaly detection algorithms for ensuring the security and efficiency of Smart City applications.
... However, as the quantity of brilliant homes increments, so are the quantity of individuals worried about their security. On account of their powerlessness to assaults and abnormal-ties, savvy homes require secure securities [1], [2]. ...
Article
Full-text available
Abnormality identification is essential in brilliant home frameworks for upgraded well-being, security, and effectiveness. This study looks at the presentation of three AI calculations-arbitrary woods, support vector machines, and strategic helping-utilizing shrewd home sensor information. Strategic supporting beat different techniques in exactness, accuracy, review, and F1 score, with better treatment of imbalanced datasets. Irregular woods succeeded in all testing means, while help vector machines had high review however lower exactness, accuracy, and F1 score. The review features the viability of strategic supporting and arbitrary backwoods for peculiarity location in shrewd homes, with potential for additional examination in this field.
... Chandak and Dharaskar [2] described a home appliance that records user's voice, "open", "close", for example. Then, it understands user's commands and performs the instructions. ...
... Speech-based user applications operated over wireless sensor networks are increasingly being utilized in various environments, for example, smart home, smart TV, and cars, since they have become a key feature of smart user interfaces [1][2][3][4]. However, as the number of the speech-based application fields has increased, the various types of background noise could negatively affect the speech sensing performance of the applications deployed over wireless sensor networks. ...
Article
Full-text available
An adaptive noise sensing method is proposed to improve the speech sensing performance of speech-based applications operated over wireless sensor networks. The proposed method is based on nonnegative matrix factorization (NMF), which consists of adaptive noise sensing and noise reduction. In other words, adaptive noise sensing is performed by adapting a priori noise basis matrix of the NMF, which is estimated from the noise signal, resulting in an adapted noise basis matrix. Subsequently, the adapted noise basis matrix is used for the NMF decomposition of noisy speech into clean speech and background noise. The estimated clean speech signal is then applied to a front-end of the speech-based applications. The performance of the proposed NMF-based noise sensing and reduction method is first evaluated by measuring the source to distortion ratio (SDR), the source to interferences ratio (SIR), and the source to artifacts ratio (SAR). In addition, the proposed method is applied to an automatic speech recognition (ASR) system, which is a typical speech-based application, and then the average word error rate (WER) of the ASR is compared with that employing either a Wiener filter, or a conventional NMF-based noise reduction method using only a priori noise basis matrix.
... and Line (http://line.me/en/) [2,3], as shown in Figure 1. ...
Article
Full-text available
With the rapid evolution of the smart home environment, the demand for natural language processing (NLP) applications on information appliances is increasing. However, it is not easy to embed NLP-based applications in information appliances because most information appliances have hardware constraints such as small memory, limited battery capacity, and restricted processing power. In this paper, we propose a lightweight morphological analysis model, which provides the first step module of NLP for many languages. To overcome hardware constraints, the proposed model modifies a well-known left-longest-match-preference (LLMP) model and simplifies a conventional hidden Markov model (HMM). In the experiments, the proposed model exhibited good performance (a response time of 0.0195 sec per sentence, a memory usage of 1.85 MB, a precision of 92%, and a recall rate of 90%) in terms of the various evaluation measures. On the basis of these experiments, we conclude that the proposed model is suitable for natural language interfaces of information appliances with many hardware limitations because it requires less memory and consumes less battery power.
Chapter
Evolution of the smart home has, to a large extent, been driven by technological developments, frequently neglecting actual human needs and habits. To realize the vision of the Wise Home, a shift to a more human-centered approach is needed. In order to move to a more natural interaction, we must first accept that current interaction paradigms are unnatural. In nature, human communication is not based on a single modality, such as speech or gesture alone, but combines them as two separate but overlapping streams of communication.
Article
A recent trend in human-computer interaction is to ease the creation of content such as apps and games via intelligent systems, allowing nonskilled users to define a given system's behavior through visual programming or simplified metalanguages. However, when the number of elements to be controlled increases, the complexity could become comparable to that of traditional coding strategies. This article addresses this issue by proposing a framework for automatically configuring a system's behavior based on user input and context information. Framework effectiveness has been tested in a game-creation scenario and used for automatically mapping user commands on virtual character actions based on a natural language description of the game scene. The use of semantics-based mapping reduces the effort and complexity linked with the configuration of the interaction logic, thus decreasing the number of commands needed for controlling the characters. The Web extra includes the questionnaire and full catalogue of descriptions collected.
Article
Full-text available
An adaptive noise sensing method is proposed to improve the speech sensing performance of speech-based applications operated over wireless sensor networks. The proposed method is based on nonnegative matrix factorization (NMF), which consists of adaptive noise sensing and noise reduction. In other words, adaptive noise sensing is performed by adapting a priori noise basis matrix of the NMF, which is estimated from the noise signal, resulting in an adapted noise basis matrix. Subsequently, the adapted noise basis matrix is used for the NMF decomposition of noisy speech into clean speech and background noise. The estimated clean speech signal is then applied to a front-end of the speech-based applications. The performance of the proposed NMF-based noise sensing and reduction method is first evaluated by measuring the source to distortion ratio (SDR), the source to interferences ratio (SIR), and the source to artifacts ratio (SAR). In addition, the proposed method is applied to an automatic speech recognition (ASR) system, which is a typical speech-based application, and then the average word error rate (WER) of the ASR is compared with that employing either a Wiener filter, or a conventional NMF-based noise reduction method using only a priori noise basis matrix.
Article
Full-text available
The performance of the sensors of a low cost Short Message System (SMS) based home security system equipped with motion sensor, smoke detector, temperature sensor, humidity sensor and light sensors has been studied. The sensors are controlled by a microprocessor PIC 18F4520 through the SMS having password. The operation of the home security has been tested on Vodafone-Fiji network for emergency and feedback responses for 25 samples. The GSM experiment showed that it takes about 8-10s for the security system to respond the occupant and relevant civil authorities in case of emergency. It takes about 18-22s for the occupant to switch and monitor lights and appliances and then get feedback from home depending upon the network traffic.
ResearchGate has not been able to resolve any references for this publication.