Abstract:A mobile network has higher demands for the performance of personalized mobile network services,but existing researches have been unable to modify the contextual mobile user preferences adaptively and providereal-time, accurate personalized mobile network services for mobile users. This paper proposes a contextcomputing-based approach to mobile user preferences adaptive learning, which can ensure the accuracy and theresponse time. First, through analyzing the logs of contextual mobile user behaviors, the method judges whethermobile user behaviors are affected by context or not, and detects whether the contextual mobile user behaviorschange. According to these judgments, the contextual mobile user preferences are modified. Secondly, the context isintroduced into the least squares support vector machine (LSSVM), which is employed to learn the changedcontextual mobile user preferences. Further, a learning method of contextual mobile user preferences is proposedwhich is based on context of the least squares support vector machine (C-LSSVM). Finally, the experimental resultsshow that the proposed method is superior to other learning methods when considering both accuracy and responsetime. The proposed method in this paper can be applied in the system of personalized mobile network services.

Abstract:To be in accordance with the thinking habits of human and improve the efficiency of reasoning, thisstudy argues to stratify weighted bases. The study shows that the existing compilation approach to non-stratifiedweighted bases can also be applied to COMPILE stratified weighted bases; however, its time and space costs arerelatively high because of redundant information in the compilation results. The paper proposes a novel compilationapproach, which can remove the redundant information in the process of compilation, and presents two optimizationtechniques to further improve the time efficiency. As with the existing approach, re-compiling a stratified weightedbase is not required whenever the weights associated with soft constraints change with time. The approach is testedby compiling random instances into ROBDD-normal bases, and the preliminary experimental results show that thetime and space costs of this approach are lower than the existing approach for most instances.

Abstract:This paper proposes the neighborhood partition graph to describe the relationship between qualitativespatial relations and actions, which is based on conceptual neighborhood graphs. The new approach is used toaddress the problem of automated planning about qualitative spatial relations. Using neighborhood partition graph,the representation and reasoning for automated planning of qualitative spatial relations is proposed. Finally, thecorrectness of the algorithm is proved with an example to describe the application. The new approach is dealsenough with dynamic qualitative spatial relations, and it has potential application in robot navigation.

Abstract:Relation extraction is a fundamental task in information extraction, which is to identify the semanticrelationships between two entities in the text. In this paper, deep belief nets (DBN), which is a classifier of acombination of several unsupervised learning networks, named RBM (restricted Boltzmann machine) and asupervised learning network named BP (back-propagation), is presented to detect and classify the relationshipsamong Chinese name entities. The RBM layers maintain as much information as possible when feature vectors aretransferred to next layer. The BP layer is trained to classify the features generated by the last RBM layer. Theexperiments are conducted on the Automatic Content Extraction 2004 dataset. This paper proves that acharacter-based feature is more suitable for Chinese relation extraction than a word-based feature. In addition, thepaper also performs a set of experiments to assess the Chinese relation extraction on different assumptions of anentity categorization feature. These experiments showed the comparison among models with correct entity types andimperfect entity type classified by DBN and without entity type. The results show that DBN is a successfulapproach in the high-dimensional-feature-space information extraction task. It outperforms state-of-the-art learningmodels such as SVM and back-propagation networks.

Abstract:Mining the latent conversations which are implied in the big amount of text messages stored on one’smobile phone, is a challenging problem. They can hardly be organized by threads, due to lack of necessary metadatasuch as “subject” and “reply-to”. This paper proposes an innovative conversation recognition model based ontemporal clustering algorithms and topic detection methods. The study first clusters the text messages into candidateconversations based on their temporal attributes, and then does further analysis using a semantic model based onlatent Dirichlet allocation (LDA). In the end, the text messages are organized as conversations based on theirintegrated correlation of temporal relevancy and topic relevancy. This approach is evaluated with a real dataset,which contain 122 359 text messages collected from 50 university students during 6 months.

Abstract:The paper presents a novel and effective heuristic algorithm for the two-dimensional rectangular strippacking problem. This algorithm is mainly based on the bestfit value and tree recursive search rules and selects themaximal fitness rectangle to the packing the space. The computational results on a large number of Benchmarkproblems have shown that this algorithm is more effective than the existing novel algorithm.

Abstract:On the basis of the traditional methods extracting information, this paper defines the formal model ofentity activity based on case grammar and presents a method based on supported vector machine and extendedcondition random fields to extract Web entity activities accurately. First, in order to automatically train the machinelearning models, the study puts forward a heuristic method to transform the semantic role labeling training data intothe training data of entity activity extraction. Next, the study trains a support vector machine classifier and extendscondition random fields using the training data. Third, using the classifier, the study distinguishes the sentences thatcontain Web entity activities. The paper also proposes forward and extends condition random fields to model thefrequency and relationship feature. The traditional conditional random fields cannot model this while the new modelcan label the entity activity information in natural language sentences more accurately. Finally, the experimentalresults show that the method is effective in multidomains and can be applied to Web entity activity extraction.

Abstract:The existing works on parsing show that lexical dependencies are helpful for phrase tree parsing.However, only first-order lexical dependencies have been employed and investigated in previous research. Thispaper proposes a novel method for employing higher-order lexical dependencies for phrase tree evaluation. Themethod is based on a parse reranking framework, which provides a constrained search space (via N-best lists orparse forests) and enables the parser to employ relatively complicated lexical dependency features. The models areevaluated on the UPenn Chinese Treebank. The highest F1 score reaches 85.74% and has outperformed allpreviously reported state-of-the-art systems. The dependency accuracy of phrase trees generated by the parser hasbeen significantly improved as well.

Abstract:The ν-support vector regression (ν-SVR) proposed by Sch lkopf, et al., has the advantage of using theparameter ν to control the number of support vectors and margin errors, however, compared to ε-SVR, itsformulation is more complicated. Until now, there have been no effective methods used to compute the ν-path for it.This paper proposes a new solution path algorithm, which is designed based on a modified formulation of ν-SVRand traces the solution path with respect to the parameter ν. Through theoretical analysis and experiments, resultscan show that the algorithm can avoid the infeasible updating path, and fit the entire ν-path in finite steps.

Abstract:With the rapid increase in size and complexity of software, more and more attention is paid to thesoftware’s trust. Verifying whether programs satisfy the properties described by assertions is a common method toguarantee trust of the software. Since path sensitive program verification cannot traverse all paths, it needs mergethe path information, which causes a loss of precision. The study proposes a program verification method usingSMT solvers, which can reduce the path search space and improve the precision at the same time. The method’smain sprit is impacting the cycle path by using maximal strongly connected component and slicing the CFGaccording to the aim assertion. The study abstracts the path space using Boolean formulas and verifies the path bycombining abstract interpretation and symbolic execution. The study has conducted experiments based on the F-Softprogram verification platform and SMT solver Z3, and results show that this method performs well based onprecision and effect.

Abstract:Web Service is a major computing resource and a main software paradigm. With rapid increase of WebServices in recent years, methods of accurately discovering the required service are becoming a research focus. Byusing service matching method, based on conception similarity, in this paper, the study transforms user requirementsand Semantic Web Service description documents OWL-S profile into ontology trees separately. Next, byhierarchical and taxonomical method, according to the ontology trees, the study computes the conception similarity,attribute similarity, and structure similarity respectively, which effectively avoids complex reason. Thereafter,according to the relationship between conception similarity and structure similarity, the study defines the sets ofconstraints and restructures the requirement trees with constraints to improve the precision and recall of the servicediscovery. Finally, the paper presents the algorithm of Semantic Web Service discovery and conducts experimentsby developing the prototype system OWLS-CSR, which proves the feasibility and affectivity of this method.

Abstract:During the construction process of an e-government project, the acquirer commonly participates in aseries of development activities handled by suppliers. These participation activities have a significant influence onthe project performance. However, when planning for participation activities that rely on subjective intuition thisraises a tense claim on policymakers’ capacity, and can arouse controversy. To this end, the study proposes analternative approach, which relies on objective knowledge from historical data. The study collects 25 completedChinese e-government projects, leverages on a variable-selection-based regression analysis, and establishes a modelof quantitative relationship between the acquirer’s participation and project performance. The model has goodmathematical properties. The study further examines the model’s degree of conformity to common experiences ofproject managers. Taking all the comparison and analysis in consideration, the study proposes several suggestionsfor acquirers in Chinese e-government projects.

Abstract:Existing distributed memory parallelizing compiler systems are mostly developed based on sharedsystems. The parallelism recognition technologies of shared memory parallelizing compiler systems are suitable forOpenMP code generation. Their implementation is used to recognize all nested loops by the same technology, sothat the parallelism cannot be efficiently explored when applying them to distributed memory parallelizing compilersystems. Thus, this paper proposes some parallelism recognition technologies suitable for the MPI code generationfor distributed memory parallelizing compiler systems by classifying the nested loops according to their structures.To solve these problems, a new classification method of nested loops is proposed, according to the structure ofnested loops and characteristics of MPI parallel program. Corresponding parallelism recognition technologies fordifferent nested loops are also presented, respectively. The experimental results show that compared with thedistributed memory parallelizing compiler systems that used existing parallelism recognition technologies, thecompiler systems, which use the proposed classification method and the corresponding recognition technologies,can more efficiently recognize parallel nested loops in the benchmark programs, and the performance speedup of theMPI codes automatically increased to more than 20%.

Abstract:The dynamic fluctuation of workload influences system metrics, affects the precision of anomalydetection. This paper proposes an online anomaly detection approach for Web applications, which handles workloadfluctuation in both request pattern and volume. The study proposes an incremental clustering algorithm to recognizeonline workload patterns automatically. For a specific workload pattern, the study adopts local outlier factor todetect anomaly and qualify the anomaly degree, and then locate the abnormal metrics with a student’s t-test method.The experimental results show that the clustering algorithm can accurately capture workload fluctuations in atypical Web application, and demonstrate that the approach is capable of not only detecting the typical faults in Webapplications, but also locating the abnormal metrics.

Abstract:Recent research in multiple DAG workflows in heterogeneous systems have been making progress andhave solved some problems, but fail to classify the multiple DAGs, according to the demand of the performanceasked by the varied DAG workflow and also fail to address the scheduling multiple DAGs workflow with multiplepriorities submitted at different times. To solve these problems, the paper presents a new model of multiple DAGsmanagement system for multiple DAGs workflow with multiple priorities and an adjustment method to the previousFairness algorithm to solve the fairness issue in scheduling multiple DAGs with the same priorities submitted atdifferent times. In addition, the study also proposes an implementation method of the Backfill algorithm formultiple DAGs with different priorities to improve utilization rate of resource, and then, based on the new modeland the two methods, propose a hybrid strategy for scheduling multiple DAGs with multiple priorities submitted atdifferent times. These experimental results show that it is possible to meet different requirements of DAGssubmitted at different times and to improve utilization rate of a resource. In addition, the results about schedulingtwo-DAGs show a significant “Trail Ending” principle, which is valuable for academic study and application.

Abstract:Reconfigurable systems, composed by many components, can exhibit different function caused bycomponent combinations and alternations. This paper focuses on the lack of methodology used in describing formalmodel and reconfigurable attributes of the reconfigurable system. The paper also focuses on the abstracts anddescribes the attributes and the behaviors of reconfigurable system and the component combination andreconfigurable method by an algebraic method. By understanding the component combination of an operation that isa new idea and extending the calculus in algebraic process, some component combination operators are defined.Next, a formal algebraic model of reconfigurable system is proposed. Based on this model, some reconfigurableattributes are analyzed and a few reconfigurable nomal formats are proposed. All above viewpoints construct thetheoretical footstone for designing reconfigurable system. Finally, a case study is introduced to explain how theabove algebraic model can be used.

Abstract:To solve leakages as complex subdivision structures, high nodes overlap probability and poorsupporting on multi-dimensional entity object retrievals and the computation of existing spatial indexes. This paperpresents a boundary restricted non-overlapping sphere tree for a unified multidimensional solid object index. Byusing the outer product expression of a sphere in Geometric Algebra, an approach is explored for intersectionjudgments and point extractions between lines and planes and lines and spherical surfaces based on meet operators.The element subdivision of multi-dimensional object voxelization and the boundary restricted non-overlappingsphere-filling algorithm is developed to balance the conditions (e.g. granularity, non-overlapping subdivision ofobject voxelization), the duplication and approximate degrees of approaching objects. Furthermore, generating andupdating minimum boundary sphere and batch Neural Gas hierarchical clustering algorithm is also presented, andcontains a volume adjusted, index level system steady with the balance of each branch of a sphere tree. Next, withthe advantage of a clear geometric meaning, simple geometric relations calculation among spheres and dynamicalupdateable parameters, index structure can be dynamically generated and updated. Finally, the unifiedmultidimensional solid object index, a query mechanism of any location and range on and in the solid objects isproposed. The updated minimum boundary sphere construction, algorithm, and the volume adjusted adaptive batchNeural Gas hierarchical cluster algorithm are defined to quickly, robustly, relatively, and uniformly classify thefilling sphere. The simulation of different physical objects and performance comparison with a commonly usedsphere tree indexes suggest that the proposed index can effectively query any position or regions on and in the solidobjects, and support the nearest linkage distance and dynamical overlapping query under limited time restrictionswith high precision.

Abstract:A novel relational routing scheme is proposed to solve the problem of querying and aggregatingsemantic media content on the network at an effective pace. The scheme can route the semantic media query andquickly backtrack relevant content from the network based on the named media and semantic association. First, themodels of semantic media and relational query are introduced, along with the necessary data structures andalgorithms for relational routing protocol. In particular, the relational matching algorithm, the procedure ofrelational routing and the approach for avoiding incomplete response of pending relational query are elaborated on.Second, the key problems of relational routing are discusse, such as media naming, query preference and theapplication. Finally, the experimentation of relational routing model and algorithms on real platform are made.Results show that the scheme is highly efficient for retrieving semantically relevant content and provides aneffective approach for semantic aggregation of distributed and dynamic media.

Abstract:The virtual network provider (VNP) cannot diagnose all service faults of virtual networks, because thesubstrate network is transparent for VNP within the network virtualization environment. To solve this problem, thepaper presents a service fault propagation model based on mapping relationships. In terms of the large fault set, thelarge symptom set, and noisy and dynamic environments, which result in the higher false positive rate and thelonger running time of existing fault diagnosis algorithms, a service fault diagnosis algorithm based on inherentcorrelation among symptoms (SFDoIC), is proposed. Simulation results show that algorithm SFDoIC can solve thedifficult problems in fault diagnosis that are caused by the transparency of the substrate network for VNP,effectively reducing the false positive rate and decreasing running time.

Abstract:Since the node’s energy dissipation model is not taken in to account in the traditional coverage models,based on the collaboration coverage model, the energy efficient hierarchical collaboration coverage model isproposed, which can evenly balance the energy dissipation among different layers in the target monitor area. Thispaper solves two specific problems in the ant colony solution. They are a formula of heuristic factor calculating forthe model and the upper and lower bounds of node numbers. Simulations in Matlab show that the proposed model ismore suitable for practical deployment which can evidently prolong the network lifetime.

Abstract:The key-exposure problem has perhaps been the most devastating attack on a cryptosystem.Conventional public key cryptosystems can revoke public keys in the case of key-exposure. However, the publickey for an attribute-based scheme represents an attribute set should not be changed. Key-Insulation is a crucialtechnique for protecting private keys. To deal with the key-exposure problems in an attribute-based cryptosystem,this paper extends the parallel key-insulated mechanism to ciphertext policy attribute-based encryption scenariosand introduces the primitive of ciphertext policy attribute-based parallel key-insulated encryption (CPABPKIE).After formalizing the definition and security notions for CPABPKIE, a concrete CPABPKIE scheme is presented.The security the proposed CPABPKIE scheme can be proved in the selective-ID model. The new primitive does notincrease the risk of helper key-exposure, and allows frequent key updating.

Abstract:Attribute revocation is crucial to use of ABE. The existing ABE schemes that support attributerevocation under the direct revocation model can only revoke the whole attributes that the user possesses byrevoking the user’s identity, so the attribute revocation is coarse-grained. This paper proposes the model of CP-ABEthat supports fully fine-grained attribute revocation. Based on the dual encryption system proposed by Waters, aconcrete CP-ABE scheme that fully supports fine-grained attribute revocation is constructed over the compositeorder bilinear groups, and the study proves its security under the standard model. Compared to the existing relatedschemes, this scheme is much more flexible and can revoke an arbitrary number of attributes that user possesses.

Abstract:The traditional ways of estimating key management schemes, which only test one, or a fewcommunications, are not comprehensive. With the increase of heterogeneous strength, static analysis will becomemore and more complex, and finally lose its referential value. First, a key management framework in this paper issummarized. According to the strategy layer of this framework and based on the classical cluster model of the entitylayer, a key management logic (KML) model is proposed. This KML model involving colored hierarchical Petri netand generalized stochastic Petri net extended the energy place. Next, by making the Top layer key managementstrategy and Button layer energy comsuption model as the core, a KML model for key management is built tosupport and represent heterogeneity with tokens to help the analysis and decisions. Finally, the performance analysismethods are discussed, including energy consumption, latency and lifetime. The experimental results show thatKML model is effective for analyzing the short-term and long-term performance and can help scheme designers toimprove their schemes by discovering the bottlenecks and hidden perils.