http://repository.iiitd.edu.in/xmlui/handle/123456789/117 2026-04-10T22:13:24Z http://repository.iiitd.edu.in/xmlui/handle/123456789/207 TeKnowBase : a tool for enrichment of textbooks using discussion forums Kongara, Amani; Bedathur, Srikanta (Advisor) Several knowledge resources are available both online and o line in learning technical topics. Textbooks act as a basic reference with their general organization into sections where each section is dedicated in explaining a single topic. Other online resources like Wikipedia articles and its topic hierarchy help the users in structured learning of a speci c technical topic by providing them with the details on advanced applications. Various discussion forums aid the users in clarifying the doubts on real world implementation details of the technical topics. For an e ective learning of technical topics, all these features are to be curated. By making use of online knowledge resources,through TeKnowBase, we present our early explorations in trying to bridge the gaps in textbooks by providing more details on the technical topic to be learnt. To extract the discussions on the details of real world implementation and advanced applications of the topics, we make use the data in StackOver ow, a discussion forum on computer programming. Extraction of relevant discussions on a speci c topic is performed using query expansion with various keywords. Two approaches are used in the query expansion, one using the keywords from Wikipedia category hierarchy and the other using the keywords describing context of a topic in textbook. A database of topics is built from the index terms and their parent topics of textbooks. Using these parent topics as keywords, we expand our search in Stackover ow for extracting more relevant discussions on the selected topic. Keywords from the above mentioned resources helps in re ning the search and extraction of relevant discussions by setting the context from the textbook to learn a topic and by using the category hierarchy of Wikipedia. The results obtained from the expanded query search are evaluated manually. Both the techniques showed an improvement over the normal keyword search in extracting relevant discussions when queried using the search framework of Lucene and evaluated using the graded evaluation measure of DCG@k. 2014-12-12T09:15:31Z http://repository.iiitd.edu.in/xmlui/handle/123456789/206 R3 : reduce, reuse and recycle Anwer, Samit; Purandare, Rahul (Advisor) Every Android application runs in its own virtual machine, with its own Linux user account and corresponding permissions. Although this ensures that permissions are given as per each application’s requirements, each permission itself is still broad enough to possible exploitation. The heap memory can be accessed by default by all apps and can be misutilized to unimaginable extents. Such exploitations may result in an over consumption of phone’s resources, in terms of memory, battery, and communication bandwidth. In this work, we propose a tool called R3, for the app developers and users to control application’s permissions at a fine granularity thereby reducing the exploitation of permissions. We provide the developers an opportunity to recycle the objects that are short lived and created in large numbers so that they can be reused instead of getting garbage collected. The framework is based on static code analysis and code instrumentation. It takes in compiled code and so does not require access to source code of the application. As a case study, we passed publicly available applications through R3 to fine tune their performance. We compared energy, data and memory consumed by these applications before and after the code injection to corroborate our claims of improvement in performance. The data consumption reduced by a factor of 12.2 after removing advertisements, energy consumption reduced by a factor of 1.88 by optimizing the wake lock type and energy consumption reduced by a factor of 3.7 after optimizing GPS location update frequency. The pause times due to garbage collection reduced from 184 ms to 80 ms as the object pool size was increased from 0 to 1000. 2014-12-12T07:12:02Z http://repository.iiitd.edu.in/xmlui/handle/123456789/205 Semantic similarity through hierarchical abstraction of knowledge Arora, Kanchan; Bedathur, Srikanta (Advisor) Identifying semantic similarity between two texts has many applications in NLP including information extraction and retrieval, word sense disambigua- tion, text summarization and type classi cation. Similarity between texts is commonly determined using a taxonomy based approach, but the limited scalability of existing taxonomies has led recent research to use Wikipedia's encyclopaedic knowledge base to nd similarity or relatedness. In this the- sis, we propose Hierarchical Semantic Analysis, a method which represents semantics of a text in high dimensional space of Wikipedia concepts and category hierarchies. We represent the meaning of any text excerpt as a weighed vector of Wikipedia-based resources. To evaluate the similarity of texts in this space, we compare the corresponding vectors using conventional metrics (e.g. cosine). Compared with the previous state of the art, use of Hierarchical Semantic Analysis(HSA) results in substantial improvements in correlation of computed similarity scores with human judgements from r= .873 to 0.901 for short sentence pairs and from r= .72 to 0.863 for paragraph pairs. 2014-12-12T07:07:16Z http://repository.iiitd.edu.in/xmlui/handle/123456789/204 SOS : save our object space Aggarwal, Aniya; Purandare, Rahul (Advisor) Each Android application (app) runs in its own virtual machine (VM), with every VM allocated a limited heap size for creating new objects. The heap size is scarce and device dependent. The more heap space an app uses, the more work the garbage collector (GC) would have; the more work the GC has, the bigger is the pause time for collection of un-referenced objects. To avoid frequent garbage collection, the objects should be allocated wisely. In this work, we propose a tool called SOS to help the developers to control and reuse memory allocated to objects on the heap. In this work, we target objects allocated in loops and identify them by leveraging static program analysis techniques. With the intention to reuse the heap space allocated to these objects, we further perform program transformation. As a case study, we take Android apps and manifest the benefits that SOS can provide in terms of reduction in pause times and reduction in heap space used. We show the trends in pause times, number of GC invocations and heap space freed, as a function of number of temporary objects. 2014-12-12T07:03:54Z