Based on the tradition safety distance models for vehicle avoidance in the braking process,the model on the safety distance is established inconsidering of the speed relationship between the front and the latervehicle,and the gradual process of vehicle deceleration.Simulation done with Matlab verified this model could resolve the inaccurate defects in computing safety left by the tradition model in theory.At last,an intersection traffic-wave simulation system is established taking Visual C++ 6.0 as software development platform,and the results.Show that the modified model can improve the efficiency of the road and reduce the traffic pollution and intersection total delayundercondition that the safe distance production is ensured.

A packet routing strategy with a tunable parameter α based on the degree and the free-degree of the node is proposed,and the data transmission on BA scale-free network is studied.A large number of numerical simulations show that when the parameter α=0.5,the BA scale-free network achievesthe maximal communication capacity,the smallest average transmission time of data packets and the smallest network load.In comparison with the local routing strategy,this strategy could improve the network capacity and reduce the average transmission time of the packet.The study is meaningful for designing the routing protocol and controlling the congestion in actual communication networks.

Finite groups all of whose n-maximal subgroups are conjugate or have the same order are investigated,and a complete classification of finite groups all of whose n-maximal subgroups are conjugate or havethe same order is given when 2≤n≤4.

By using the theories of the Deskins index complex ofthe maximal subgroup of finite groups,some relational theorems in the relevantreference are continued,and two new criteria of the solvable groups are obtained.

A class of sharp Hardy type inequalityfor generalized Baouendi-Grushin operators is presented by improving the methods of the Euclid space.Then,a class of Hardy-Sobolev type inequality is established by its results.As applications,the positive property and the unbounded property from below for p-degenerate subelliptic operator are discussed.

This paper considers cellular neural networks (CNNs)with impulses and delays.Some new sufficient conditions ensuring the existence and exponential stability of anti-periodic solutions for CNNs are established.

This paper mainly describes state equivalence of finite recognizers and its properties.A finite recognizer is a unique decomposable finite recognizer to exist in the sense of isomorphism.It also gives the state simplification and algorithm of state equivalence classes on finite recognizers.

The antioxidant activities of petroleum ether extract(PE),chloroform extract (CE),ethyl acetate extract (AE),ethanolic extract (EE),and alkaloids ofUncaria rhynchophylla (Miq.)Jacks.were determined with flow-injection chemiluminescence and compared with Vitamin C (Vc).The results indicated that all extracts exhibit excellent antioxidant activities.The antioxidant activities were in the sequence of EE＞Vc＞AE ＞CE＞Alk＞PE.

A new fluorescein-based fluorescent derivatizating reagent,6-oxy-(isopropionyl ethylenediamine) fluorescein (IPEF),was synthesized and characterized by MS and ¹H NMR.The fluorescence spectra and varietyof IPEF and its valeric acid-derivative were measured respectively.The fluorescence intensity and photostability of IPEF and its valericacid derivative were assessed in comparison with fluorescein.The results indicated that the intensities of the new one increase with the increase of pH and its intensity in methanol is a littleweaker than in water but much stronger than that in acetonitrile.

Sarafloxacin·3H₂O was characterized by X-ray single crystal diffraction analysis.It belongs to monoclinic crystal system,space group is P-2₁/n with cell parameters:C₂₀H₁₇F₂N₃O₃,M_r=385.36,a=1.473 1(3),b=0.864 24(17),c=1.535 9(3) nm,α=90°,β=96.785(6)°,γ=90°,V=1.941 8(7) nm³,Z=4,D_c=1.472 Mg/m³,F(000)=900,μ=0.120mm^-1,R₁=0.121 8,wR₂=0.322 7 (I＞2σ(I)).A two-dimensional network isgenerated by hydrogen bonds.

The predominance area diagrams of lgPS₂-lgPO₂ and lgPS₂-lgPSO₂ for Fe-S2-O2 were drawn by the improved component activity term method of base element.And the influences on the purity of pyrite by the residual oxygen impurity inthe process of synthesis have been characterized according to these drawn predominance area diagrams.The results showed that:(1)While the pyrite is preparedby heat-treatment of iron or iron oxides in flowing high-purity inert gas withgaseous sulfur to ensure the purity of pyrite,the residual oxygen must be removed by complete reaction of oxygen with sulfur before synthesis pyrite.(2)The high-purity pyrite can be prepared by heat-treatment of iron or ironoxides with sulfur in sealed quartz reactor.

Physiological characters of different restoration tree species in NongLa peak cluster depression were mensurated by LI-6400,such as Radermachera sinica,Toona sinensis,Bambusa lapidea,Zenia insignis and Leacaena leucacephala.By comparing their daily change of net photosynthetic rate P_n,and solar radiation PAR,etc,it can be found that environment factors daily changes were complex.Communities of different restoration tree speciesadjust to the microclimate differently.Zenia insignis and Radermachera sinica were in higher average P_n level.Leacaena leucacephala showed “midday depression” of photosynthesis obviously.Radermachera sinica was in higher average E level.Leacaena leucacephala and Zenia insignis were inhigher average WUE level.Toona sinensis was in the highest average AQE level.Different factors influenced photosynthesis of these restoration treespecies differently.Stepwise regression method was used to establish linear regression equations of net photosynthetic rate and various factors,which were statistically significant and had a good preditability.

Heat shock proteins of plants were main protein in some evolution stress.Expression of some heat shock proteins had a certain relationship with plant tolerance.The nucleotide sequence of a cDNA cloned from Arabidopsis thaliana ecotype RLD was determined,and the corresponding amino sequence deduced.The open reading frame encodes a DnaJ protein of 442 residues with a molecular mass of 47 ku.The predicted sequence contains regions homologous to theJ-and cysteine-rich domains of Escherichia coli DnaJ.The cDNA fragment directionally cloned into the vector pET32α,over-expression of DnaJ in E.coli,the E.coli could grow on LB containing 0.5 mol/L NaCl.

The experiments were carried out to find out subcellular distribution characteristics and chemical forms of Mn in the leaves,stems and roots of Polygonum pubescens Blume,which is a newly found manganese hyperaccumulator.Differential centrifugation and sequential chemical extraction were used.More than 90% of total Mn was bound to cell wall and soluble fraction in theplant.The results from extraction experiments by using different extractants showed that water-extractable Mn and NaCl-extractable Mn predominated in stems,while water-extractable Mn,NaCl-extractable Mn and HCl-extractable Mn predominated in roots and leaves.With the Mn supply increasing,the percentage of Water-extractable Mn decreased while the percentage of HCl-extractable and NaCl-extractable Mn increased in the roots,and Water-extractable and HCl-extractable Mn increased in the leaves.Most of the Mn,which was accumulated by P.pubescens Blume,wasstored in the cell wall and the soluble fractions (including cytoplasm and karyon).For the chemical forms,Water,HCl and NaCl-extractable Mn predominated in the plant.

30 healthy Tibetan piglets with similar body weight(6of suckling Tibetan piglets of 1 d,7 d,14 d,21 d and 28 d,respectically) are chosen for detecting developmental expression of Pept1,EAAC1 and CAT1 mRNA through real time PCR.Results show that there are greater significance for ontogenesises of Pept1,EAAC1 and CAT1 mRNA expression along the horizontal axis of the small intestine of suckling Tibetal piglets.Ontogenesises of Pept1 and CAT1 mRNA expressionin duodenum are similar,while 14-day-old expression of the highest abundance,compared with other developmental stages,was significant(P＜0.01);The expression of EAAC1,CAT1 mRNA developmental changes are similar in Jejunum,the expression of CAT1 in the 21-day-old Tibetan piglets reach the highestpoint,but compared with others no significant difference;Ontogenesises of Pept1 and CAT1 mRNAexpression in ileum are similar,Pept1 and CAT1 mRNA expression abudance both are highest at 7-day-old.

To investigate the effect of rhFGF-8a to chick embryo chorioallantoic membrane angiogenesis.RhFGF-8a protein is over-expressed in E.coli through genetic engineering technology and insoluble components is about 50% of the total protein.RhFGF-8a protein is refoled andpurified and its purity can reach 99% identified by SDS-PAGE.The biologicalactivity of rhFGF-8a is determined by MTT and its effect on angiogenesis is identified by chick embryo chorioallantoic membrane(CAM) experiments.The results showthat the purified rhFGF-8a protein can stimulate the growth of NIH 3T3 cells and rhFGF-8aprotein has weak capacity on angiogenesis in CAM.

As the k-nearest neighbor imputation (kNNI)algorithm is often biased in choosing the k nearest neighbors of missing data,a new imputation method is put forward,Quadrant-Encapsidated-Nearest-Neighbor based Imputation method (QENNI),for missing values.The algorithm uses the quadrant nearest neighbors (points of the encapsulant) around a missing datum to impute the missing datum.It is not biased in selecting nearest neighbors.Experiments demonstrate that QENNI is much better than the kNNI method in imputed accuracy.

Association rules are one of the new reserch methodsfor statistical data analysis of English corpus.Based on analysis of English teachers' oral corpus,association rules are applied to study cognitive process of teachers' explicit representation of tacit knowledge and students' tacit knowledge structure fromthe viewpoints of mother tongue,questions,and directive words respectively.Thestatistical experimental results prove its effectiveness and feasibility.

This paper briefly outlines the deficiencies of the main existing datamodels in the Semantic Web environment.Based on the inherent characteristics of the academic content,a Structured Digital Object (SDO) data model and the definition of SDO are given.Taking academic journalsas examples,this paper introduces an academic information retrieval system SDO-based,its design and functional development.Finally,conclusions and expectation forthe future works are presented.

The ontological question answering system needs to convert the questions in natural language into ontology query statements,and the present solutions are mainly natural language interface and question similarity.To overcome the disadvantage of sentence similarity applied in ontological QA system,the improved method of computing similarity is designed.Via building the set of query pattern for frequently asked questions,the similarity of target question iscomputed based on the statistic,semantic and structure of question.The query pattern of target question is chosen in automatic and interactive mode separately,then it is converted into real SPARQL query statements,and finally ontology is retrieved to extract answers.The accuracy rates of the two modes are 83.8% and 92.1% separately.

Concept lattice is the core of Formal Concept Analysis (FCA).A novel algorithm of constructing concept lattice based on evolutionismis proposed.First,the initial seed concepts come into being by the single attribute of a given formal context.Then the variable seed concepts and stable seedconcepts are chosen respectively.Child stable seed concepts and child variableseed concepts are generated from stable seed concepts sets.And then the new generations are produced by the child stable seed concepts and child variable seedconcepts.Follow the circle until no new concept is generated.The algorithm explores the views of the evolutionism which makes a vivid constructing concept lattice process and have higher efficiency.

By introducing the ideas of k-means clustering,genetic algorithm andminimum vertex cover,a global optimal police patrol car configuration and deployment for the department of traffic police is proposed.In this setting,k-meansclustering is used to generate N center points based on the density of the intersections and completed the police patrol car initialization configuration.Thenthe genetic algorithm is adopted to receive a global optimal patrol route as reference and finally the Dijkstra algorithm is employed to calculate the patrol deployment routes.Meanwhile,the calculation methods of the shortest path is given between any two intersections and the location of police cars at anymoments,as well as regional coverage and travel time of the N police patrol cars.Toverify its validity,sensitivity analysis on the budget,the number of patrol cars is performed to identify the possible solutions.Experimental results show this method is robust and can effectively improve the effectiveness of patrol deployment and the routes kept varied with good property of concealment.

A new mutlsurfaceproximal SVM,called Multi-surface Proximal Support Vector Machine via Maximum One-class ScatterDeference Combination (SCGEPSVM) is proposed.This approach firstly modifies theprimal problem of GEPSVM,then incorporates One-class scatter into it.With comparable computational cost to GEPSVM,SCGEPSVM overcomes many deficiencies.Experiments carried out on toy and publicly available datasets which disclose its effectiveness.

The design principles of Nutch's plug-in mechanism are discussed.Nutch's lightweight plug-in architecture includes three basic concepts.They are micro-kernel,extension point and extension;plug-in,plug-in manifest and plug-in manifest descriptor;plug-in factory and plug-in manifest parser.The processes of plug-in and the mappings between relative concepts and real files in Nutch are also summarized.It's obvious that in-depth understanding of Nutch's plug-in mechanism can contribute to build flexible new applications based on Nutch.

It is very urgent to improve abilities of current search engines.This paper,firstly proposed a novel approach for personalized Web search through seamless integration with mainstream search engines—a model of search engine-browser plugin-recommendation engine;secondly designed a framework of Web search recommendation system based on case-based reasoning guided by above model;thirdly,discussed the acquirement,presentation and organization of search experiences and proposed a primary structure of case model.Finally introduced a Websearch recommendation system based on this model—ExpertRec.

This paper proposes a personalized recommendation model based on results clustering analysis.First,a user model is established accordingtothe user behavior;Then the search results are re-sorted by using multiple information fusion algorithms.Finally clustering analysis on results is carried out and recommendations are made by adjusting the weight of different class document.The experimental results of teaching scientific research information recommendation show that this model can improve the recommendationeffectiveness.

In order to improve the efficiency and concurrency and avoid single point of failure of the FTP SE,a novel FTP searching engine is presented based on the distributed and P2P technology,which is named “Large Scale FTP searchingengine based on P2P and distributed technology”.This paper discussed three partsof the SE:information collection,retrieve and index modules,and primary focus on the distributed information collection scheme and the DHT-based inverted index algorithm.

Based on the analysis of removing duplicate strategies in collecting Web information,which are used by the Bloom Filter algorithm and its improved versions and combining with Dynamic Bloom Filter algorithm,this adopts dynamic arrayto represent the elements of aggregate,and then proposes a removing duplicate strategy,which supports frequently querying and deletes operation of repeated URL.Finally,an experiment is carried out by using the proposed strategy,and comparing it withother strategies,which shows that the strategy gets better effect in removing duplicate in the case of lower error rates.

An automatic generating method of adaptable Web information extraction rule is proposed,which extracts data of complex type on the Web andrevises iteration by comparing with multi-sampling page and discovering repeatpattern in one-sample and extracts dynamic base on method of semantic analysisrules.Experimental results show that the method performs well in both precision and robustness.

Routing mechanisms is one of key parts in knowledge discovery.The paper proposes a mixed P2P semantic Routing retrievalmodel based on Ontology Knowledge Query Tree.User node can match search term with knowledge of OKQT by querying OKQT on routing nodes,and makes sure the location query fleetly and accurately,and then queries the knowledge which meet user precisely.For the search term which is not match on the internet,by calculating the semanticsimilarity between search term and knowledge in OKQT,and zooming the search term,the corresponding results are given back.Experiment verifies that this method which make sure the location first and query knowledge second can improve the semantic routingefficiency,recall ratio,precision ratio and expansibility.

A new literature retrieval system is built to returnthe summary of a paper based on its literature impact.It can also returnthe comments given by other papers.The summary and the comments help readers quickly understand both the value of the paper and its inadequacy,whichmay not be found in the paper's abstract.The comments found in citing papers are virtualized as “query” generated by uni-gram language model,and thesentences of original paper are treated as a set of “document”.Using KL-divergence scoring approach to find the similarity between the query and the documents,the top-scored impact sentences of the original paper are selected and returned as summary.The corresponding comments in the citing papers are also returned.The System isbased on the Platform for Applying,Researching And Developing Intelligent Search Engine (PARADISE) developed by Peking University.It has the advantage of quick-to-start and good scalability.

In order to improve theaccuracy of the corpus of complete parsing,this paper analyzes the corpus which was amended by people and summarizes the reason leading to the annotation inconsistency.Moreover,some strategies to eliminate inconsistencies and explored disambiguation type are proposed from various hierarchies such as the word segmentation,Pos tagging,and parsing structure.Experiments show that the method described in this article can improve the accuracy of corpus annotation by 2.5%.

Based on consumer psychology and sentiment analysis,using consumptionmotivation type theory,fuzzy modeling for the evaluation and emotion of onlinereview texts is proposed and corresponding fuzzy corpus is established.A calculation method of comprehensive evaluation and emotion with respect to the consumer's preference for product attributes is obtained.Furthermore,values of evaluation and emotion are regarded as reasoning antecedents.The establishmentof fuzzy inference rules recommendation to consumers of four different motivations is received as the consequent.The validity of the proposed method is proved by experimental results.

With the popularity of Web applications,Web contentsecurity has beenaroused great attention.Web content security monitoring and classifying have become a key research issue.Based on the analysis of Web content security,demandedmodel is proposed in this paper.The model,combined with VSM,employs the Vagueset to improve the feature extraction strategy on the set of original documentrepresentation model.The results of the experiment on the medium-sized Web pages fromthe real Web corpus show that this model can improve the Web content security text classification results,compared with the traditional one.

Term extraction is a basic subject in information processing and is attracting more and more attention nowadays.In order to extractlow frequency words effectively,Log-likelihood ratio method is used but with alow precision rate.To solve this problem,C-value method is used to deal withthe results of Log-likelihood ratio.Experiment results show that by combining the two methods,the precision is improved in the premise of ensuing high recall rate of Log-likelihood ratio method.The proposed method can improve the precision by about8% compared with the Log-likelihood ratio method.

Nowadays,with the trend of language diversity inthe internet,how to organize multi-language resources becomes a hotspot.This paper focuses on cross language text categorization (CLTC) which can organize heterogeneous document collections.Using the semantics pairs extracted from parallel.The latent semantic dual space can be built by using the semantics pairs extracted from parallel corpus.In the experiment,through changing the trainingsize and language composition the performance of CLTC can be verified.Results show that cross-language text classification based on latent semantic dual space makes a good performance in stability and accuracy.

This paper presents a machine learning algorithm which combines the semantic and statistical aspect fortext sentiment classification.First,extract the text of the emotional wordsas features,and use the statistical methods to get the initial weight,then re-weight the feature by analyzing the semantic structure of the text;finally use Bayesian algorithms and Bayesian classification algorithm as a basic classification of the Boosting algorithm to classify the text.The empirical resultsshow that,the accuracy of the Bayesian classification algorithm based on semantic understanding is higher than the Bayesian classification algorithm based on statistic and its accuracy can reach 90%.

In the useof Fourier transform phase correlation image alignment principles of right,based on a improved traditional Fourier transform image registration algorithm canbe improved by using the pyramid from coarse to fine hierarchical registration strategy,and high robustness the image with accuracy can be effectively improved.Experimental results show that the algorithm has quasi-precision,high speed,and highly robust.

An improved sub-pixel offset algorithm is studied.Setting tangent threshold,the gradient direction of edge is determined by theratio and the positive or negative sign of partial differential.Then it inhibits pseudo-edgeof the identified single-pixel wide edge points in gradient image.Finally,it carries on the quadratic parabolic interpolation of gradient amplitude function alongits gradient direction,and calculate sub-pixel offset.Experimental results show that this algorithm has short running time and high detection accuracy and canbe applied to the industrial automation online fast image measurement tasks while in the condition of not very high precision.

Siraidia grosvenrii natural varieties and fruit image of Tissue-cultured ones are processed and calculated with self-developed Siraidia grosvenriiimage recognition system to gain various species of the appearance of thecharacteristic parameters,such as length,width,eccentricity,perimeter,area,circular degree,solidity,red,green,blue hue and saturation,etc.Then the feature information database of Siraidia grosvenrii varieties is built,which helpto detect Siraidia grosvenrii natural variety,clear their category boundaries of characteristics,and offer threshold selection reference.Therefore,thisdatabase is the foundation of machine vision inspection and identification of Siraidia grosvenrii varieties.