Ontogator:Combining View-and Ontology-Based
Search with Semantic Browsing
Eero Hyv¨o nen,Samppa Saarela,and Kim Viljanen
Helsinki Institute for Information Technology(HIIT)/University of Helsinki
P.O.Box26,00014UNIV.OF HELSINKI,FINLAND
Eero.Hyvonen,Samppa.Saarela,Kim.Viljanen@cs.Helsinki.FI
http://www.cs.helsinki.fi/group/seco/
In:Proceedings of XML Finland2003,Open Standards,XML,and the Public Sec-tor,Kuopio,October30-31,2003.
Abstract.We show how the bene?ts of the view-based search method,devel-
oped within the information retrieval community,can be combined and extended
with the bene?ts of ontology-based annotations and search,developed within the
Semantic Web community.As a proof of the concept,we have implemented an
ontology-and view-based image retrieval and recommendation browser Ontoga-
tor.Ontogator is innovative in two ways.Firstly,the RDFS-based ontologies used
for annotating image metadata are used in the end user interface to facilitate view-
based image retrieval.The views provide the user with a search function and
means for getting useful overviews of the contents in the repository.Secondly,
a semantic browsing function is provided by a recommender system.This sys-
tem enriches instance level image metadata by the ontology and provides the user
with links to semantically related relevant images.The notion of a“semantically
relevant link”is speci?ed in terms of logical rules.To illustrate and discuss the
ideas,a practical application of Ontogator to a photo repository of the Helsinki
University Museum is presented.
1Introduction
There are two major approaches to image information retrieval.In content-based image retrieval(CBIR)[3]the images are retrieved based on their characteristics,such as color,texture,shape,etc.In the metadata-based approach to be discussed in this paper, image retrieval is based on descriptions of the images.
To retrieve images from a database,keyword-based query systems[1,2]are typi-cally used.Here the user may select?ltering values or apply keywords to the different database?elds,such as the“creator”,“time”,or to the content descriptions including classi?cations and free text documentation.More complex queries can be formulated, e.g.,by using Boolean logic.
Keyword annotations and keyword-based information retrieval systems are widely used but have several problems related1)to the quality of search results and2)to the usage of systems.
1.1Answer Quality Problems
The precision and recall of keyword-based search methods is lowered due to many reasons[4].For example,a keyword in a document does not necessarily mean that the document is relevant,and relevant documents may not contain the explicit word. Synonyms lower recall rate,homonyms lower precision rate,and semantic relations such as hyponymy,meronymy,and antonymy[6]are not taken into account of.
A prominent solution approach to problem is to use ontology-based annotations and information retrieval[16,17].
1.2Usability Problems
The standard keyword search methods are not always easy to use[9]:
Formulating the information need Keyword search implicitly assumes that the user has a goal in mind,i.e.,to?nd a set of images with desired characteristics.However, in many applications this is not the case.One may,for example,want to learn about some topic and only have a general interest of?nding images related to a vague topic.
Formulating the query The user is often faced with a repository of images whose content is semantically complicated and the domain more or less unknown.In such situations it is dif?cult to to determine what keywords to use in formulating the query.
Formulating the result set The answer format used in a keyword-based search system is typically a list of hits,the result set,ordered according to their expected relevance to the user.Explanations on why different hits are included the result,and/or their semantical grouping would be helpful in analyzing the results,too.
A solution approach to address the usability issues above is the multi-faceted or view-based search method1[14,8].Here the idea is to organize the terminological key-words of the underlying database into orthogonal hierarchies and use them extensively in the user interface in helping the user to formulate the queries,in navigating the database,and in grouping the results semantically.
The traditional notion of hit lists and view-based search misses an important aspect of the repository content:the relations by which the hit items(images)are related with each other and other images in the database.This relational information should in many cases be a part of the answer as,for example,recommendations.For example,if the query contains the keyword“Sibelius”,and the result set contains an image depicting Jean Sibelius,the Finnish composer of symphonies inspired by the Carelian scenery(a part of Finland),then a relation to images of Carelia(not in the actual result set)could be of interest to the user.
This paper describes a system called Ontogator.Its main novelty lays in the idea of enhancing keyword search accuracy and usability by combining ontology-based knowledge representation with the view-based search method.Furthermore,the notion
of view-based searching is complemented with the idea of semantic browsing used in Topic Maps[12]and recommender systems[15].
We describe Ontogator by using a concrete application case in which the system has been developed.However,the idea of Ontogator is not bound to any particular application domain.A goal of the work is to develop a generic semantic image browser for image repositories,whose contents are annotated by associating each images with a set of semantic RDF(S)resources describing its content.
In the following,keyword,view-based and ontology-based approaches to image retrieval are?rst discussed.After this the Ontogator system is presented and its under-lying ideas are discussed.In conclusion,the lessons learned and contributions of this work are summarized.
2Semantic Image Annotation and Retrieval
2.1Keyword-Based Annotation and Retrieval Schemes
The content of images in a database are typically described by associating each image with a set of keywords that describe its content.The keywords can be either explicitly user-given or be determined automatically from free text and other descriptions of the images.
Keywords are often selected from controlled vocabularies or thesauri[7]in order to maintain annotations mutually coherent and to ease image retrieval.Since controlled thesauri are not complete and new keyword terms emerge,also to use of uncontrolled vocabularies is often necessary.
Additional expressive power to keyword annotations can be obtained by hierarchical thesauri and classi?cation systems,such as the Art and Architecture Thesaurus[13]or ICONCLASS2[18].They classify different aspects of life into hierarchical categories.
A category is described by its name and a set of additional keywords.When an image is annotated by a category C,the image automatically inherits the keywords of C and its super-categories.For example,since“castle”is a subcategory of“building”,an im-age annotated with the keyword“castle”is found using the keyword“building”.The same idea of enlarging a keyword with related terms in order enhance recall is used in thesaurus-and concept-based query expansion techniques[10].
2.2View-Based Search Method
The thesaurus can be constructed in a systematic fashion by a set of semantically or-thogonal hierarchies such as“Time”,“Place”,etc.that are often called facets or views. The facets provide complementary views of the contents along different dimensions.
The facets can be used for indexing the content and to help the user during infor-mation retrieval[14].Firstly,the hierarchies give the user an overview of what kind of information there is in the repository.Secondly,the hierarchies can guide the user in creating the query and in selecting appropriate keywords that are likely to lead to non-empty answer sets—a recurring problem in IR systems.Thirdly,the hierarchies can
be used to disambiguate query terms.Fourthly,the facets can be used as a navigational aid when browsing the database content[8].
The idea in multi-facet search is that the user makes a selection of categories of
interest from different facets,and the system then constructs the corresponding query. If the categories selected are C1C n and the subcategories of C i i1n,including itself are S i1S i2S i k,respectively,then this corresponds to the following boolean
AND-OR-query:
S11S1k S21S2l S n1S n m(1) An additional idea of the multi-facet search is to compute proactively the number of hits in every direct subcategory of C i i1n and show it to the user.In this way,the user can be hindered from making a selection leading to empty result set and be guided toward such selections that are likely to constrain or relax the search appropriately.
This idea was used,e.g.,in the HiBrowse system[14]in the90’s.A later application
of the multi-facet approach is the Flamenco system[8],the?rst web-based proto of the Ontogator system[9]and its current version described in this paper.In Flamenco,the next level of subcategories for the selected categories are exposed to user.View-based search is adapted into a navigational browsing-searching scheme for the Web.The query is constrained further during the navigation?ow by clicking subcategory links from the next direct subcategory level below,or by relaxing formerly selected constraints.After each step,next selection possibilities and the sizes of the corresponding result sets are computed.Extensive user studies[11,5]have recently been carried out to show that a direct Google-like keyword search interface preferred if the users know precisely what they want.However,if this is not the case,then the multi-faceted search method with its“browsing the shelves”sensation is clearly preferred over keyword search or using only a single facet.
In Ontogator,the search interface is based on the HiBrowse model.However,the whole hierarchy,not only the next level of subcategories,can be opened for selections. Moving between hierarchy levels is more?exible because at any point any new se-lection in the hierarchy opened is possible.The“browsing the shelves”sensation is provided by a separate recommendation system based in the underlying ontological domain knowledge.This provides a semantically richer basis for browsing than the keyword hierarchies used in Flamenco.
2.3Ontology-Based Annotation and Search
If view-based search is based on keywords then the search method does not solve the answer quality problem of keyword-based search,which is due to the fact that a set of keyword cannot describe accurately the contents of images.For more accurate descrip-tions,semantically richer ontology-based annotations can be employed[16,17].Such annotations are not atomic keywords but can be more detailed structured descriptions that are linked with other resources in a semantic graph.With the help of ontologies, the user can also express the queries more precisely and unambiguously,which leads to better precision and recall rates.Furthermore,through ontological class de?nitions and inference mechanisms,such as property inheritance,instance-level metadata can be
automatically enriched and used for determining implicit semantic relationships in the database.The price to be paid is that much more work is needed when constructing the ontologies and during the content annotation https://www.doczj.com/doc/048145482.html,ing more complex and detailed ontological structures in the user interface may also make the system less understand-able to the end-user.
3Ontogator Approach
The key idea of the Ontogator system is to combine the usage bene?ts of multi-facet search with the answer quality bene?ts of ontology-based search,together with seman-tic recommendations.To describe the system,we use the promotion ceremony image database of the Helsinki University Museum3as a case study.Promotion ceremonies consist of several academic occasions and parties and last for several days.The database contains629photographs about the ceremony events and documents,ranging from the 17th to21th century,and more images are acquired after every promotion.The prob-lem is to provide the museum guest with a sensational information retrieval system for investigating the contents of this semantically complicated image database,i.e.,to illustrate the inner life and traditions of the university.
Fig.1.Architecture of Ontogator.
Figure1depicts the overall architecture of Ontogator.The system is used by the Content Browser and is based on two information sources:Domain Knowledge and Annotation Data.
Domain Knowledge consists of an ontology that de?nes the domain concepts and the individuals.In our case,the domain ontology consists of some329promotion-related concepts,such as“Person”and“Building”,125properties,and2890in-stances,such as“Linus Torvalds”and the“Entrance of Cathedral of Helsinki”.
Annotation Data describes the metadata of the images represented in terms of the an-notation and domain ontologies.Annotation ontology describes the metadata struc-ture used in describing the images.It is assumed,that the subject of an image is described by associating the image with a set of RDF(S)resources of the domain knowledge,classes or instances.They occur in the image and hence characterize its content.The difference with simple keyword annotations is that the associated resources are part of the domain ontology knowledge base,which disambiguates the meaning of annotations(synonym/homonym problem)and provides additional implicit semantic metadata.The annotations also include other metadata,such as the photographer,free text descriptions and some technical information of the im-ages,but in the promotion case application only the subject descriptions are used for multi-facet search and recommendations.
Based on the domain knowledge and the annotation data,Ontogator provides the user with two services:
Multi-facet search The underlying domain ontologies are mapped into facets and fa-cilitate multi-facet search.In our example case,there are six facets“Happenings”,“Promotions”,“Performances”,“Persons and roles”,“Physical objects”,and“Places”.
The facets provide different views into the promotion concepts and data and are used by the user to focus the information need and to formulate the queries,as described earlier.
Recommendation system After?nding an image of interest by multi-facet search,the domain ontology model together with image annotation data can used to recom-mend the user to view other related images.The recommendations are based on the semantic relations between the selected image and other images in the reposi-tory.Such images are not necessarily included in the answer set of the multi-facet search query.For example,images of the next and previous events in the promo-tion ceremonies can be recommended or images of the relatives of a person in the ?gure.The recommendation system makes it possible to browse the contents using semantic navigation.
The two services are connected with the information sources by tree sets of con?g-urations or rules.
Hierarchy rules The hearth of the multi-facet search engine is a set of category hi-erarchies by which the user expresses the queries.The hierarchy rules are a set of con?gurational rules that tell how to construct the facet hierarchies from the domain ontologies.
Mapping rules Annotations associate each image with a set of resources of the domain ontology.Mapping rules extend these direct annotations by describing the indirect relations between the images and the domain knowledge.For example,a direct image annotation may tell that a particular person,say Linus Torvalds,is in an image.However,the person may be in different images in different roles,e.g.,as a “Promovend”or as a“Doctor Honoris Causa”.Roles are a useful facet to the user of the system.An image in which Linus Torvals is present in the role of Doctor Honoris Causa should therefore be annotated with this role in order to distinguish
the image from other images of him.However,then the image is not annotated directly as an image of Linus Torvals and the view-based search would not?nd the image as an image of the person.Mapping rules solve the problem by specifying the relations by which images are related with domain resources. Recommendation rules The domain ontology de?nes not only the concepts and their hierarchical structure,but also the relations by which the actual domain classes and individuals are related with each other.Based on these relations,recommendation rules are used to?nd associations between an image and other images of poten-tial interest to the user.The recommendations are de?ned in terms of logical Horn clauses.For example,“Related Person”-rule may link a person with another per-sons through family relations.If the user selects an image exposing a person p,then images exposing persons in different family relations with p can be recommended to the user.
4View-Based Search Method
The view-based search makes faceted hierarchical categories used to describe the image content visible to the user.Hierarchies may represent,for example,places,happenings, time,roles and persons.Figure2shows the search interface of the Ontogator prototype. The query is formulated and executed by selecting resources of interest from the facet hierarchies.When the user makes a selection,the system retrieves the images that are related to the selected resource.When several resources are selected from different views,the result is the intersection of the images related to these resources(cf.section 2.2).After each selection the result set is recomputed.For each resource in the opened hierarchies,a number ratio n k is counted and shown.It tells that if the resource is selected,then there will be n images in the result set out of the k images related to that resource totally.A selection leading to empty result set(n0)is disabled and shown in gray color.
The inner nodes of the hierarchies are associated with their sub-nodes by the search system.If the facet hierarchy is projected using rdfs:subClassOf and rdfs:type relations,and C is the selected class,then the system retrieves images that are either related to the class C directly or to any of its subclasses or instances.
The underlying hierarchies can also be used to formulating the results of a query. For example,Ontogator constructs descriptions of images by listing the resources of the selected categories that actually appear in a picture and shows them beside the thumbnail pictures.Another possibility would be to group the results according to the subcategories of a user-selectable facet[8].
4.1Hierarchy Rules
Hierarchy rules de?ne the root categories of the facets and how the facet hierarchies are projected starting from these.Hierarchy rules are needed in order to make the classi?-cations shown to the end user independent from the design choices of the underlying Domain Ontologies.The view-based search system itself does not differentiate between differently projected hierarchies.
S e l ec t
e d r e s ou r ce s
D i s a b l e d r e s ou r ce s
a r e g r a y e d ou t
D e s c r i p ti o n o f t h e
s e l ec t e d r e s ou r ce
ca n b e v i e w e d Fig.2.Ontogator user interface for view-based search.
An obvious way to extract a facet hierarchy from the RDF(S)-based domain knowl-edge is to use the subclass-of hyponymy relation.Then the inner nodes of the hierarchy consist of the classes of the domain ontology,and the leaves are the direct instances of these https://www.doczj.com/doc/048145482.html,ing only hyponymy for facet projections would,however,be a lim-itation in the general case.For example,places may constitute a meronymical part-of hierarchy,and this would a natural choice for a facet in the user interface.
If the hierarchies intersect each other,then a resource selection should be considered in the context of the hierarchy.For example,Helsinki may be viewed as an instance of the class City,a legal body,or as a part of Finland.Choosing Helsinki from the part-of hierarchy should match pictures of squares,beaches and other places that are situated in Helsinki,but it would be confusing,if pictures of beaches were returned by a query where Helsinki is selected in the sense of a legal body.
The idea of viewing an RDF(S)knowledge base along different hierarchical projec-tions has been applied,e.g.,in the ontology editor Prot′e g′e -20004that allows to choose the property by which the hierarchy of classes shown to the user is projected.When using the default hierarchy constructed by the subClassOf -relation,the root of the hi-erarchy is always the top class (”Thing”in Prot′e g′e -2000,”Resource”in RDF Schema).When using some other relation,a problem is how to determine the root since,for exam-ple,there may be cycles in the RDF graph.In Prot′e g′e -2000the selected class becomes
the root of the hierarchy.Prot′e g′e-2000also has a general option”References”that uses any instance-valued property to construct the hierarchy.This idea could be applied in Ontogator as well.However,in many cases a more precise speci?cation of the projec-tion than a single property is needed.For example,the hyponymy projection already employs two properties(rdfs:subClassOf and rdf:type).Furthermore,the ordering of the sub-resources may be relevant.In our case,for example,the sub-happenings of an event should be presented in the order in which they take place.In the Ontogator pro-totype,the projections are created by special purpose Java methods implementing the hierarchy rules,and only hyponymy projections were used.Ordering of the sub-nodes can be speci?ed by a con?gurable property.
Hierarchy rules tell how the hierarchies used in the view-based search are projected.
A closely related but still separate question is how these hierarchies should be shown to the user.For example,in our case study,the ontology was created partly before the actual annotation work and had more classes and details than were actually needed. The projected Objects facet hierarchy,for instance,had many classes of decorations not related to any picture.A hierarchy may also have intermediate classes that may be useful for knowledge representation purposes but are not very natural categories to the end user.One needs to be able to?lter unnecessary resources away from the user interface,yet they should be present internally in the search hierarchies.In our work, con?gurational?ltering rules for showing the facets have been investigated but not yet implemented in the system.
4.2Mapping Rules
An image is annotated by associating it with a set of domain knowledge resources.This set is,however,not enough because there may be complex indirect relations between images and resources describing its content.Mapping rules are used to specify what indirect resources describe the images in addition to the direct ones.Through such rules it is possible to achieve a search system that is independent of both the annotation scheme and the domain ontology design.The search system itself does not make any distinction between the ways in which resources and images may be related.
For example,there are the classes Role and Person in the domain ontology of pro-motions.The subclasses of Role,such as Master and Doctor Honoris Causa,are used to indicate the role in which some person may appear in a picture.If the role r of a person p appearing in a picture is known,then the picture is annotated by an instance of the Role.As a result,the picture is found using r in the multi-facet search,but not with p,which is unsatisfactory.The system should be able to infer that the images,that are about an instance of Role are also images about the person in that role.
A similar kind of situation occurs with the concept of promotion.All instances of the class Promotion are related to a particular university,faculty,and the conferrer of degrees of the promotion.The system should be able to infer that pictures related to some promotion are also related to the university and faculty of that particular promo-tion happening.However,in contrast,the conferrer of degrees is related to the image only if actually appearing in it.This kind of distinctions are highly domain dependent and dif?cult to?nd out automatically without explicit additional information,such as
mapping rules.The mappings between annotations and resources of domain knowledge can be quite complex.
In Ontogator,mapping rules are given as RDQL5query templates.The templates are applied for hierarchy resources r,classes and instances,by?rst instantiating them with the URI of r.The resulting RDQL query is applied to the RDF(S)knowledge base consisting of the Domain Ontology and the Annotation Data.The result is a set of image resources related to r.All mappings between facet resources and the images are determined when constructing the system’s inner representation of the facet hierar-chies.This strategy of computing mappings during the startup makes the search system faster but at the price of the memory needed for the search data structures.In the future versions of Ontogator,we intent to develop a dynamic version that uses the underlying RDF representation directly through a general inference engine.
The applicability of a mapping rule is usually limited only to a certain subtree of the facet hierarchy.Furthermore,if facet hierarchies intersect,then the mapping rules may be facet-dependent.
5Rule-Based Recommendations
In addition to the multi-facet search mechanism,Ontogator also has a rule-based rec-ommendation utility.It allows the user to browse semantically related images in the spirit of Topic Maps[12].However,while the links in a Topic Map are given by the map,the links in Ontogator are inferred based on rules and the underlying knowledge base.
Figure3illustrates the recommender system.The Query overview lists the selected facet categories and Query results the result set.The Recommendations for the Selected picture are grouped on the right based on three rules:one for determining pictures de-picting related persons(family relations),one for determining pictures of the preceding event,and one for determining pictures of the following event.
An RDF(S)knowledge base contains relations between the resources described in the ontologies and the metadata.Not all of the relations and resources may be of interest of the user,and it may also be undesirable to show all information.(For example,the limited monitor size of a mobile device constrains the amount of information to be shown.)Furthermore,a related resource may not be interesting in itself,but only as a mediating resource for another resource.In such cases,the user would be interested in knowing directly the indirect resource behind the mediating resource.
Recommendation rules are needed for selecting the most interesting relations from the wealth of relations between resources in the knowledge base.The relations may be either direct property links or indirect ones.Through recommendation relations the end user gets a more?uent,semantically guided browsing experience between resources interest.
Q u e r y r e s u lt s S e l ec t e d p i c t u r e
R ec o mm e nd a ti o n s
Fig.3.Screenshot of the recommendation system.
5.1Alternatives for Recommendations
Recommendations can be created in various ways[15].In our case we have been considering following three alternatives:user pro?le-based,similarity-based,and rule-based recommendations.
User pro?le-based recommendations are based on information collected by observ-ing the user,or in some cases by asking the user to explicitly de?ne the interest pro?le. Based on the user’s pro?le,recommendations are then made to the user either by com-paring the user’s pro?le to other users’pro?les(collaborative?ltering/recommending) or by comparing the user’s pro?le to the underlying document collection(content-based recommending).
The strength of user pro?le-based recommendations is that they are personalized and hence serving better the user’s individual goals.In our case application,personal-ization is however dif?cult,because the users cannot be identi?ed.It is not even known when the user’s session begins and when it ends because the users are using the same physical kiosk interface located in the museum.The pro?ling must be easy for the user because most of the users use the system perhaps only once in their lifetime.Finally, it is dif?cult to identify whether the user liked or disliked the current image without asking the user to rate every image explicitly.A weakness could be also that explaining the recommendations to the user can be dif?cult,because they are based on heuristic
measures of the similarity between user pro?les and database contents,and of the user’s actions.
With similarity-based recommendations we refer to the possibility to compare the semantical distance between the metadata of a selected image and the other images.The nearest images are likely to be of more interest and could be recommended to the user.
A dif?culty of this recommendation method is how to measure the semantical distance between metadata.In many cases the most similar image is not the most interesting one but rather just another picture of the same event.A simple method is to use the count of common or intersecting annotation resources as a distance measure,but there are lots of other choices based on the ontology available.
The idea of rule-based recommendations used in Ontogator is that the domain spe-cialist explicitly describes the notion of”interesting related image”with generic rules. The system then applies the rules to the underlying knowledge base in order to?nd in-teresting images related to the selected one.This method has several strengths.Firstly, the rule can be associated with a label,such as”Images of the previous event”,that can be used as the explanation for the recommendations found.It is also possible to deduce the explanation label as a side effect of applying the rule.Recommendation rules are described by the domain specialist.The rules and explanations are explicitly de?ned, not based on heuristic measures,which could be dif?cult to understand and motivate. Secondly,the specialist knows the domain and may promote the most important rela-tions between the images.However,this could also be a weakness if the user’s goals and the specialists thoughts about what is important do not match,and the user is not inter-ested in the recommendations.Thirdly,the rule-based recommendations do not exclude the possibility of using other recommendation methods.For example,the recommen-dation rules could perhaps be learned by observing the users actions and then used in recommending images for the current or future users.
In the?rst web-based version of Ontogator[9],we implemented pro?le-based and similarity-based recommendation system that recommended more semantically similar images.The recommendation were not static but were modi?ed dynamically by main-taining a user pro?le and a history log of image selections.Then a rule-based recom-mendation system was implemented due to the bene?ts discussed above.The idea was tested in the promotion application that currently contains1)rules for recommending images of related persons,such as children,parents,wife,husband,etc.,and2)rules for determining images of the next and previous events based on the general promotion programme used in(almost)all promotions.
5.2Implementation
The current version of the rule-based recommender is implemented in SWI-Prolog6. For reasons of ef?ciency,the recommendations are determined in a batch process before using the application.The recommender reads the metadata and ontologies in RDF(S) format into the Prolog interpreter.The semantic relations between the images are then determined by the rules.Finally,the program produces a XML-?le containing the rec-ommendations which is loaded into the Ontogator browser at the next startup of the
browser.A limitation of this static approach is that it is not possible to create on-line
dynamic recommendations based on the user’s pro?le and usage of the system.
The Prolog rules are divided in three groups:domain speci?c recommendation rules, system speci?c rules(the“main”program creating the recommendations)and RDF
Schema speci?c rules(such as rules implementing the transitive subclass-of closure). The domain speci?c rules are created by the domain specialist.System speci?c rules
and RDF Schema speci?c rules are domain independent and need to modi?es only if
the system or the RDF Schema speci?cation changes.
When processing the data,the program iterates through all images and their meta-
data.The recommendation rules are applied to every different resource r in the im-
ages’metadata(a URI reference).If recommendations are found,they are stored as recommendations for r.The recommendations are created for each metadata resource
r and not for each image in order to minimize the size of the XML-?le.The Ontogator browser then shows as the recommendations of an image the recommendations related
to each resource r used in the image’s metadata.
The recommendations contain the recommended resource(URI),a natural language explanation for the relation between the resources,and a category for the recommen-
dation(e.g.,“related persons”).In the current implementation,the natural language
descriptions are relatively simple such as“Person A-a child of-Person B”.The texts are based on the labels of the resources de?ned in the RDF descriptions.Also
the reverse relation is described,which would be in the previous example“Person B
-a parent of-Person A”.This facilitates symmetric usage of the recommendation associations.
In the promotion application,the recommendation system is working as planned and creates recommendations shown to the user.The recommendations extend Ontogator usage from just searching images to browsing between related images.However,eval-uating the quality and relevance of recommendations can only be based on the user’s opinions.In our case,only a small informal a user study of has been conducted using the personnel of the museum.The general conclusion was that the idea seems useful in practice.
Logic programming seems to be a very?exible and effective way to handle RDF(S)
data by querying and inferring when compared with RDF query languages,such as RDQL and RQL.The de?nition of the recommendation rules requires programming skills and may be dif?cult to a domain specialists who is not familiar with logic https://www.doczj.com/doc/048145482.html,putational ef?ciency and central memory requirements can be a problem if the RDF knowledge base is large and if the rules are complex.
In the domain model of the promotion application,the properties did not constitute
hierarchies.However,property hierarchies would make it easier to create recommen-dation rules in situations where a recommendation rule could be described using the top level property and be“inherited”to its sub-properties.For example,if there is a property“human-relation”,the sub-properties could include“spouseOf”,“parentOf”, and“childOf”.The de?nition of a“human-relation”recommendation rule could then be done using the“human-relation”property and be(automatically)available also,e.g., for determining the persons with the spouseOf relation.
Observing the end users using the system could give valuable additional informa-tion about what recommendations are mostly used and the?ow of viewing the images. The value of such studies is,however,limited by fact that that the ontologies and cur-rent recommendations limit the possibilities of the user to select any image from the collection as the next image.
6Conclusions
This paper developed a method for combining the bene?ts of RDF(S)-based knowledge representation,the view-based search method,and knowledge-based recommendations. An implementation prototype,Ontogator,and an application of it to a practical image retrieval problem was discussed.Ontologies and facet hierarchies can be used to help the user in formulating the information need and the corresponding query.Furthermore, the ontology-enriched knowledge base of image metadata can be applied to constructing a more meaningful answer to a query than a simple hit-list.The recommender system can provide the user with a semantic browsing facility between semantically related images.
The integration of the view-based and ontology-based search paradigms turned out to be more complicated than expected.The main dif?culty is how to model and deal with the indirect relations between the images and domain ontology resources,and how to project the facet hierarchies from the RDF knowledge base.If not properly modeled the precision and recall rates of the system are lowered.
A reason for choosing RDF(S)is its domain independent nature an opennes.This makes it possible to apply Ontogator more easily to other image repositories by recon-?gurations.During our work,we actually reused the promotion ontology and instance data easily in another application.
De?ning new logical recommendation rules on top of the RDF-triple format is?exi-ble,but required a fair amount of expertise concerning the underlying ontological struc-tures and Prolog programming.A problem encountered there is how to test and verify that the recommendations for all images are feasible without having to browse through the whole database.
During our work,Prot′e g′e-2000was used as the ontology editor.Jena’s7basic main memory-based model(ModelMem)was employed to load the RDFS-models into On-togator’s internal representation form.Prot′e g′e turned out to be a versatile tool with an intuitive user interface that even for a non-programmer could use for constructing ontologies(from the technical viewpoint).A good thing about Prot′e g′e is that it is not limited to RDFS semantics only,but enables and enforces the use of additional features.
Ontology evolution poses a problem with Protege even in the simple case that a name(label)of some class changes.Prot′e g′e derives URI’s of the classes from their names,and if a name changes then the classes URI(ID)changes also.This leads to con?gurational problems.Rules and mappings for one version of the ontology do not apply to the new version,even though the actual classes have not changed,only their labels.Multi-instantiations would have been desirable in some situations but this is not possible with Prot′e g′e.
The major dif?culty in the ontology-based approach is the extra work needed in creating the ontology and the detailed annotations.We believe,however,that in many applications—such as in our case problem—this price is justi?ed due to the better accuracy obtained in information retrieval and to the new semantic browsing facilities offered to the end-user.The trade-off between annotation work and quality of infor-mation retrieval can be balanced by using less detailed ontologies and annotations,if needed.
7Acknowledgements
Kati Hein¨a mies and Jaana Tegelberg of the Helsinki University Museum and Avril Styr-man provided us with the actual case database,the ontology,and annotated the images. Our work was partly funded by the National Technology Agency Tekes,Nokia,TietoE-nator,the Espoo City Museum,the Foundation of the Helsinki University Museum,the National Board of Antiquities,and the Antikvaria-group.
References
1.M.Agosti and A.Smeaton,https://www.doczj.com/doc/048145482.html,rmation retrieval and hypertext.Kluwer,New York,
1996.
2.R.Baeza-Yates and B.Ribeiro-Neto.Modern Information Retrieval.Addison-Wesley,New
York,1999.
3.J.P.Eakins.Automatic image content retrieval—are we getting anywhere?pages123–135.
De Montfort University,May1996.
4. D.Fensel(ed.).The semantic web and its languages.IEEE Intelligence Systems,Nov/Dec
2000.
5.J.English,M.Hearst,R.Sinha,K.Swearingen,and K.-P.Lee.Flexible search and navigation
using faceted metadata.Technical report,University of Berkeley,School of Information Management and Systems,2003.Submitted for publication.
6. C.Fellbaum,editor.WordNet.An electronic lexical database.The MIT Press,Cambridge,
Massachusetts,2001.
7. D.J.Foskett.Thesaurus.In Encyclopaedia of Library and Information Science,Volume30,
pages416–462.Marcel Dekker,New York,1980.
8.M.Hearst,A.Elliott,J.English,R.Sinha,K.Swearingen,,and K.-P.Lee.Finding the?ow
in weg site search.CACM,45(9):42–49,2002.
9. E.Hyv¨onen,S.Kettula,V.Raatikka,S.Saarela,and Kim Viljanen.Semantic interoper-
ability on the web.Case Finnish Museums Online.Number2002-03in HIIT Publications, pages41–53.Helsinki Institute for Information Technology(HIIT),Helsinki,Finland,2002.
http://www.hiit.?.
10.K.J¨arvelin,J.Kek¨al¨ainen,and T.Niemi.ExpansionTool:Concept-based query expansion
and https://www.doczj.com/doc/048145482.html,rmation Retrieval,4(3/4):231–255,2001.
11.K.-P.Lee,K.Swearingen,K.Li,and M.Hearst.Faceted metadata for image search and
browsing.Proceedings of CHI2003,April5-10,Fort Lauderdale,USA.Association for Computing Machinery(ACM),USA,2003.
12.Steve Pepper.The TAO of Topic Maps.In Proceedings of XML Europe2000,Paris,France,
2000.https://www.doczj.com/doc/048145482.html,/topicmaps/materials/rdf.html.
13.T.Peterson.Introduction to the Art and Architecture Thesaurus,1994.
https://www.doczj.com/doc/048145482.html,.
14. A.S.Pollitt.The key role of classi?cation and indexing in view-based searching.Technical
report,University of Hudders?eld,UK,1998.http://www.i?https://www.doczj.com/doc/048145482.html,/IV/i?a63/63polst.pdf. 15.J.Ben Schafer,Joseph A.Konstan,and John Riedl.E-commerce recommendation applica-
tions.Data Mining and Knowledge Discovery,5(1/2):115–153,2001.
16. A.T.Schreiber,B.Dubbeldam,J.Wielemaker,and B.J.Wielinga.Ontology-based photo
annotation.IEEE Intelligent Systems,16:66–74,May/June2001.
17.G.Schreiber,I.Blok,D.Carlier,W.van Gent,J.Hokstam,and U.Roos.A mini-experiment
in semantic annotation.In I.Horrocks and J.Hendler,editors,The Semantic Web–ISWC 2002.First international semantic web conference,number LNCS2342,pages404–408.
Springer–Verlag,Berlin,2002.
18.J.van den Berg.Subject retrieval in pictorial information systems.In Proceedings of the
18th international congress of historical sciences,Montreal,Canada,pages21–29,1995.
http://www.iconclass.nl/texts/history05.html.
应用ADS设计微带线带通滤波器 1、微带带通微带线的基本知识 微波带通滤波器是应用广泛、结构类型繁多的微波滤波器，但适合微带结构的带通滤波器结构就不是那么多了，这是由于微带线本身的局限性，因为微带结构是个平面电路，中心导带必须制作在一个平面基片上，这样所有的具有串联短截线的滤波器都不能用微带结构来实现；其次在微带结构中短路端不易实现和精确控制，因而所有具有短路短截线和谐振器的滤波器也不太适合于微带结构。 微带线带通滤波器的电路结构的主要形式有5种： 1、电容间隙耦合滤波器 带宽较窄，在微波低端上显得太长，不够紧凑，在2GHz以上有辐射损耗。 2、平行耦合微带线带通滤波器 窄带滤波器，有5％到25％的相对带宽，能够精确设计，常为人们所乐用。但其在微波低端显得过长，结构不够紧凑；在频带较宽时耦合间隙较小，实现比较困难。 3、发夹线带通滤波器 把耦合微带线谐振器折迭成发夹形式而成。这种滤波器由于容易激起表面波，性能不够理想，故常把它与耦合谐振器混合来用，以防止表面波的直接耦合。这种滤波器的精确设计较难。
4、1/4波长短路短截线滤波器 5、半波长开路短截线滤波器 下面主要介绍平行耦合微带线带通滤波器的设计，这里只对其整个设计过程和方法进行简单的介绍。 2、平行耦合线微带带通滤波器 平行耦合线微带带通滤波器是由几节半波长谐振器组合而成的，它不要求对地连接，结构简单，易于实现，是一种应用广泛的滤波器。整个电路可以印制在很薄的介质基片上(可以簿到1mm以下)，故其横截面尺寸比波导、同轴线结构的小得多；其纵向尺寸虽和工作波长可以比拟，但采用高介电常数的介质基片，使线上的波长比自由空间小了几倍，同样可以减小；此外，整个微带电路元件共用接地板，只需由导体带条构成电路图形，结构大为紧凑，从而大大减小了体积和重量。 关于平行耦合线微带带通滤波器的设计方法，已有不少资料予以介绍。但是，在设计过程中发现，到目前为止所查阅到的各种文献，还没有一种能够做到准确设计。在经典的工程设计中，为避免繁杂的运算，一般只采用简化公式并查阅图表，这就造成较大的误差。而使用电子计算机进行辅助设计时，则可以力求数学模型精确，而不追求过分的简化。基于实际设计的需要，我对于平行耦合线微带
《Access Treeview 应用实例》 使用说明： 本程序可在各种支持VBA语言的软件上直接使用，其中MS Access已成功测试具体步骤为： 1、在Access里创建一个tblTreeview的数据表如下图： 2、在Access里创建一个frmTreeview的窗体，在窗体里创建一个Treeview1的Treeview控件和一个 ImageList1的ImageList控件如下图：
3、然后把程序复制到frmTreeview窗体的加载事件当中如下图： 4、最终效果如下图：
程序： Private Sub Form_Load() On Error Resume Next Dim strParent As String, strChild As String, MyNode As Node Me.Treeview1.Nodes.Clear '清除Treeview的所有旧节点 Me.Treeview1.ImageList =Me.ImageList1.Object '把图标加载到每个节点前面 Set rst = CurrentDb.OpenRecordset("SELECT * FROM tblTreeview", dbOpenSnapshot, dbReadOnly) 'Treeview 节点数据来源 While Not rst.EOF strChild = rst!ChildID If rst!ChildID <> rst!ParentID Then strParent = Mid(rst!ParentID, 1, Len(rst!ParentID) - Len(rst!ChildID) - 1) Set MyNode = Me.Treeview1.Nodes.Add(strParent, tvwChild, rst!ParentID, rst!ChildID, "A1", "A3") '加载子节点 Else strParent = rst!ChildID Set MyNode = Me.Treeview1.Nodes.Add(, , strChild, rst!ChildID, "A1", "A3") '加载父节点End If rst.MoveNext Wend rst.Close '关闭数据集 Me.Treeview1.HideSelection = False '离开焦点后有阴影 End Sub
实验八 有源滤波器的设计 一．实验目的 1． 学习有源滤波器的设计方法。 2． 掌握有源滤波器的安装与调试方法。 3． 了解电阻、电容和Q 值对滤波器性能的影响。 二．预习要求 1． 根据滤波器的技术指标要求，选用滤波器电路，计算电路中各元件的数值。设计出 满足技术指标要求的滤波器。 2. 根据设计与计算的结果，写出设计报告。 3. 制定出实验方案，选择实验用的仪器设备。 三．设计方法 有源滤波器的形式有好几种，下面只介绍具有巴特沃斯响应的二阶滤波器的设计。 巴特沃斯低通滤波器的幅频特性为： n c uo u A j A 21)(??? ? ??+= ωωω ， n=1，2，3，. . . （1） 写成： n c uo u A j A 211) (??? ? ??+=ωωω （2） )(ωj A u 其中A uo 为通带内的电压放大倍数，ωC A uo 为截止角频率，n 称为滤波器的阶。从（2） 式中可知，当ω=0时，（2）式有最大值1； 0.707A uo ω=ωC 时，（2）式等于0.707，即A u 衰减了3dB ；n 取得越大，随着ω的增加，滤波器的输出电压衰减越快，滤波器的幅频特性越接近于理想特性。如图1所示。ω 当 ω>>ωC 时， n c uo u A j A ??? ? ??≈ωωω1 )( （3） 图1低通滤波器的幅频特性曲线
两边取对数，得： lg 20c uo u n A j A ωω ωlg 20)(-≈ （4） 此时阻带衰减速率为： -20ndB/十倍频或-6ndB/倍频，该式称为衰减估算式。 表1列出了归一化的、n 为1 ~ 8阶的巴特沃斯低通滤波器传递函数的分母多项式。 在表1的归一化巴特沃斯低通滤波器传递函数的分母多项式中，S L = c s ω，ωC 是低通 滤波器的截止频率。 对于一阶低通滤波器，其传递函数： c c uo u s A s A ωω+= )( （5） 归一化的传递函数： 1 )(+= L uo L u s A s A （6） 对于二阶低通滤波器，其传递函数：2 22)(c c c uo u s Q s A s A ωωω++ = （7） 归一化后的传递函数： 1 1)(2 ++= L L uo L u s Q s A s A （8） 由表1可以看出，任何高阶滤波器都可由一阶和二阶滤波器级联而成。对于n 为偶数的高阶滤波器，可以由2n 节二阶滤波器级联而成；而n 为奇数的高阶滤波器可以由2 1-n 节二
VB TreeView控件使用详解 (2012-06-16 15:08:16) 标签：分类： 第一小时：学习直接用代码将数据填充到树控件中。 为什么要先学习直接用代码将数据填充到树控件中因为这种方法是最简单的，代码也最容易理解，学习树控件，先将这个学会，已经掌握了一半，所以先不要急着想怎么将表中的数据填充到树控件中，在第一小时里，树控件和表完全没有关系。 目的：我们要在树控件中建立如下的一个3层级关系 水果 | |__苹果 | |__红富士 | |__国光 | |__葡萄 |__红提子 |__青提子 解释：水果包含2种，一种是苹果，一种是葡萄，苹果又包含2种，一种是红富士，一种是国光，葡萄也如此。 在这里：“爷”是水果，“父”是苹果，葡萄，“子”是红富士，国光，红提子，青提子。概括如下： 爷（只能有一个）：水果 父（这里有2个）：父1：苹果；父2：葡萄
子（这里有4个）：子1：红富士（父1苹果的子）；子2：国光（父1苹果的子）；子3：红提子（父2葡萄的子）；子4：青提子（父2葡萄的子） 1、新建一个窗体，在窗体上放置两个控件，一个是Treeview，一个是Imagelist 如何找到这两个控件 Treeview控件在“工具箱”的榔头加扳手图标（其他控件）中选“Microsoft Treeview Control,Version " Imagelist控件在“工具箱”的榔头加扳手图标（其他控件）中选“Microsoft Imagelist Control,Version " Treeview控件大家都明白干什么用的，Imagelist控件是干什么用呢原来这个控件是放图标用的，如果你想在树控件中显示图标的，这个图标都将储存在ImageList控件中。 2、设置这两个控件的属性 首先要讲清楚控件的属性设置有2种，一种是设置这个控件在ACCESS中的属性，比如名称等。一种是设置这个控件本身的属性。要设置这个控件在ACCESS中的属性，选中控件后按鼠标右键选“属性”就可以了。跟我们平时设置文本框什么的一样。要设置这个控件本身的属性，只要双击这个控件就可以了。 1）设置Treeview控件在ACCESS中的名称属性，将名称设置为“Treeview" 2）设置Imagelist控件在ACCESS中的名称属性，将名称设置为“Image" 2)设置Imagelist控件本身的属性，双击控件后，在弹出来的设置框中选“Images",单击“Insert Picture"按钮，在你电脑中选择你需要的图标。在“Key:”栏中填入“K1”。其他默认设置不用改。3）设置Treeview控件本身的属性，双击控件后，在弹出来的设置框中选“General”，在这个选项面版中有很多项设置，大多数是设置树控件的显示格式，你自己慢慢研究。这里我们将第一项“Style"选7，在第五项“Imagelist"选项中将我们放置的Imagelist控件“Image"选上。这项设置就将图标和树控件联系了起来。 3、写代码将数据填充到树控件中
滤波器是一种只传输指定频段信号，抑制其它频段信号的电路。 滤波器分为无源滤波器与有源滤波器两种： ①无源滤波器: 由电感L、电容C及电阻R等无源元件组成 ②有源滤波器: 一般由集成运放与RC网络构成，它具有体积小、性能稳定等优点，同时，由于集成运放的增益和输入阻抗都很高，输出阻抗很低，故有源滤波器还兼有放大与缓冲作用。 利用有源滤波器可以突出有用频率的信号，衰减无用频率的信号，抑制干扰和噪声，以达到提高信噪比或选频的目的，因而有源滤波器被广泛应用于通信、测量及控制技术中的小信号处理。 从功能来上有源滤波器分为： 低通滤波器（LPF）、高通滤波器（HPF）、 带通滤波器（BPF）、带阻滤波器（BEF）、 全通滤波器（APF）。 其中前四种滤波器间互有联系，LPF与HPF间互为对偶关系。当LPF的通带截止频率高于HPF的通带截止频率时，将LPF与HPF相串联，就构成了BPF，而LPF与HPF并联，就构成BEF。在实用电子电路中，还可能同时采用几种不同型式的滤波电路。滤波电路的主要性能指标有通带电压放大倍数AVP、通带截止频率fP及阻尼系数Q等。 带通滤波器（BPF） （a）电路图(b)幅频特性 图1 压控电压源二阶带通滤波器 工作原理：这种滤波器的作用是只允许在某一个通频带范围内的信号通过，而比通频带下限频率低和比上限频率高的信号均加以衰减或抑制。典型的带通滤波器可以从二阶低通滤波器中将其中一级改成高通而成。如图1（a）所示。 电路性能参数 通带增益 中心频率 通带宽度 选择性 此电路的优点是改变Rf和R4的比例就可改变频宽而不影响中心频率。 例．要求设计一个有源二阶带通滤波器，指标要求为： 通带中心频率 通带中心频率处的电压放大倍数： 带宽： 设计步骤： 1）选用图2电路。 2）该电路的传输函数： 品质因数： 通带的中心角频率： 通带中心角频率处的电压放大倍数： 取，则：
带通滤波器设计步骤 1、根据需求选择合适的低通滤波器原型 2、把带通滤波器带宽作为低通滤波器的截止频率，根据抑制点的频率距离带通滤波器中心频点距离的两倍作为需要抑制的频率，换算抑制频率与截止频率的比值，得出m 的值，然后根据m 值选择低通滤波器的原型参数值。 滤波器的时域特性 任何信号通过滤波器都会产生时延。Bessel filter 是特殊的滤波器在于对于通带内的所有频率而言，引入的时延都是恒定的。这就意味着相对于输入，输出信号的相位变化与工作的频率是成比例的。而其他类型的滤波器（如Butterworth, Chebyshev,inverse Chebyshev,and Causer ）在输出信号中引入的相位变化与频率不成比例。相位随频率变化的速率称之为群延迟（group delay ）。群延迟随滤波器级数的增加而增加。 模拟滤波器的归一化 归一化的滤波器是通带截止频率为w=1radian/s, 也就是1/2πHz 或约0.159Hz 。这主要是因为电抗元件在1弧度的时候，描述比较简单，XL=L, XC=1/C ，计算也可以大大简化。归一化的无源滤波器的特征阻抗为1欧姆。归一化的理由就是简化计算。 Bessel filter 特征：通带平坦，阻带具有微小的起伏。阻带的衰减相对缓慢，直到原理截止频率高次谐波点的地方。原理截止频率点的衰减具有的经验公式为n*6dB/octave ，其中，n 表示滤波器的阶数，octave 表示是频率的加倍。例如，3阶滤波器，将有18dB/octave 的衰减变化。正是由于在截止频率的缓慢变化，使得它有较好的时域响应。 Bessel 响应的本质截止频率是在与能够给出1s 延迟的点，这个点依赖于滤波器的阶数。 逆切比雪夫LPF 原型参数计算公式（Inverse Chebyshev filter parameters calculate equiations ） ) (cosh )(cosh 11Ω=--Cn n 其中 1101.0-=A Cn , A 为抑制频率点的衰减值，以dB 为单位；Ω为抑制频率与截止频率的比值 例：假设LPF 的3dB 截止频率为10Hz,在15Hz 的频点需要抑制20dB,则有： 95.91020*1.0==Cn ；Ω=15/10=1.5 1.39624.0988.2) 5.1(cosh )95.9(cosh 11===--n ,因此，滤波器的阶数至少应该为4
树形结构控件TreeView TreeView是一种 能以树形目录结构形式 显示数据的高级控件， 显示方式类似于 Windows的资源管理 器，能分层展开各结点 的子目录，也能收缩各 结点的子目录。本节将 先介绍TreeView控件 的一些基本概念、属性 与方法，再举例说明 TreeView控件的应用。 TreeView控件概述、属性与方法 1、作用：用于显示Node结点的分层列表。 2、添加到控件箱 菜单命令：工程 | 部件，在部件对话框中选择：Microsoft Windows Common Controls 6.0 3、TreeView控件的属性 （1）属性对话框 用鼠标右键单击TreeView 控件，在弹出式菜单中选择属 性，进入属性设置对话框，该 对话框分为通用、字体、图片 三个选项卡，如图8.6所示。
①样式（Style）：返回或设置在Node结点之间显示的线样式，如表8.6所示。 ②鼠标指针(MousePoint)：可选择不同鼠标样式，如表8.1所示。 ③线条样式（LineStyle）：0-tvwTreeLine 无根结点的树形结构，1-tvwRootLines有根结点的树形结构。 ④标签编辑(LabelEdit)：0-tvwAutomatic 自动，1-tvwManual 手工； ⑤图像列表(ImageList)：结点图标所用ImageList控件； ⑥边框样式(BorderStyle)：0-ccNone无边框，1-ccFixedSingle单边框； ⑦外观（Appearence）：0-ccFlat平面效果，1-cc3D 3D效果； ⑧缩进：父子结点的水平间距。 （2）其它属性 ①SelectedItem.Text属性：用于返回或设置当前Node结点的内容。 ②CheckBoxes属性：该属性只能取逻辑值，若取True值，则每个Node结点前出现一个复选框，否则不出现复选框。 ４、TreeView控件的方法 （1）Node结点 ①Node结点：是TreeView控件中的一项，它包含图像与文本。 ②Nodes结点集合：包含一个或多个Node结点。 （2）Add方法 ①作用：为TreeView控件添加节点和子节点。 ②定义格式 TreeView1.nodes.Add(Relative,Relationship,Key,Text, Image,SelectedImage) 其中： ◆Relative参数：添加新结点时，其父结点键值Key。添加根结点时，此项为空。 ◆Relationship参数：新结点的相对位置： tvwlast—1：新节点位于同级别所有节点之后； tvwNext—2：新节点位于当前节点之后； tvwPrevious—3：新节点位于当前节点之前； tvwChild—4：新节点成为当前节点的子节点。 ◆Key：Node结点关键字（唯一标识符），用于检索Node结点。同时也作为其新建子结点的Relative值，即新建子结点的Relative=父结点Key。 ◆Text：Node结点文本。 ◆Image：Node结点位图，是关联ImageList控件中位图的索引。
DELPHI中利用TreeView控件建立目录树2000-06-26 00:00:00·－·中国计算机报社 p>Rainbow的话：关于TreeView的使用，还可以参看：联合使用TreeView 组件 TreeView是一个显示树型结构的控件，通过它能够方便地管理和显示具有层次结构的信息，是Windows应用程序的基本控件之一。DELPHI虽然具有比较强大的文件管理功能，提供了多个用于文件管理的标准控件，如DriveComboBox、DirectoryListBox、FileListBox等，通过设置它们的属性，使其建立起联系，甚至不用编写一行程序，我们就可以实现在不同的目录之间进行切换，然而这样的目录切换只适用于进行文件的查找定位，而不能方便地进行目录的浏览，例如我们要从c:\windows目录转到c:\program files目录，就必须返回到根目录才能进行切换，而不能象Windows资源管理器那样任意地在不同的目录之间进行浏览与切换。 要实现在不同目录之间任意切换和浏览，还是需要使用TreeView控件，以下程序就利用DELPHI的TreeView控件来建立目录树。 在该程序中采用的各部件以及界面设计如下图所示： 各部件的主要属性设置如下： 部件属性属性值form name caption form1 ‘目录浏览’ drivecommbobox name visible drivecommbobox1 false filelistbox name visible filetype filelistbox1 false fddirectory imagelist name imagelist1 treeview name images 该程序利用DriveCommboBox控件来获得系统具有的驱动器，并以此作为目录树的最上层，利用FileListBox控件，通过设置其Filetype属性为fdDirectory，可以获得所需的子目录，在TreeView控件的OnExpanding事件中将得到的子目录加到该控件的某一节点下。
RC 有源带通滤波器的设计 滤波器的功能是让一定频率范围内的信号通过，而将此频率范围之外的信号加以抑制或使其急剧衰 减。当干 扰信号与有用信号不在同一频率范围之内，可使用滤波器有效的抑制干扰。 用LC 网络组成的无源滤波器在低频范围内有体积重量大，价格昂贵和衰减大等缺点，而用集成运放 和RC 网络组成的有源滤波器则比较适用于低频，此外，它还具有一定的增益，且因输入与输出之间有良 好的隔离而便于级联。由于大多数反映生理信息的光电信号具有频率低、幅度小、易受干扰等特点，因而 RC 有源滤波器普遍应用于光电弱信号检测电路中。 一.技术指标 总增益为1 ; 通带频率范围为 300Hz —3000Hz ,通带内允许的最大波动为 -1db —+1db ; 阻带边缘频率范围为 225Hz 和4000Hz 、阻带内最小衰减为 20db ; 二?设计过程 1 .采用低通-高通级联实现带通滤波器； 将带通滤波器的技术指标分成低通滤波器和高通滤波器两个独立的技术指标，分别设计出低通滤波器 和高通 滤波器，再级联即得带通滤波器。 低通滤波器的技术指标为： f PH = 3000Hz A max - 1d B G =1 f SH = 4000Hz A min = 20dB 高通滤波器的技术指标为： f pL = 300Hz A max = 1d B G = 1 f si_ - 225Hz A min - 20dB 2. 选用切比雪夫逼近方式计算阶数 (1).低通滤波器阶数 N >ch 4[J(10 0.1Amin -1)/(10 0.1Ami N 1 _ ■ 1 Ch ( f SH / f PH ) (2).高通滤波器阶数 N 2 ch'[ *. (10 0.1Amin -1)/(100.1Amax -1)] Ch^(f pL /f SL ) 3. 求滤波器的传递函数 1) .根据Ni 查表求出归一化低通滤波器传递函数 H LP (S)二 H LP (S)| S S' 2= --- 2冗PH 2) .根据Na 查表求出归一化高通滤波器传递函数 N 2 H_P (S '，去归一化得 H^s '，去归一化得
TreeView 控件显示Node 对象的分层列表，每个Node 对象均由一个标签和一个可选的位图组成。 本文主要介绍C# treeView控件中，添加，修改、删除节点等c# treeview控件的使用方法。 其代码如下： 1.private void Form1_Load(object sender, EventArgs e) 2.{ 3. https://www.doczj.com/doc/048145482.html,belEdit = true;//可编辑状态。 4. 5.，这个结点是根节点。 6. TreeNode node = new TreeNode(); 7. node.Text = "hope"; 8. treeView1.Nodes.Add(node); 9. TreeNode node1 = new TreeNode(); 10. node1.Text = "hopeone"; 11. TreeNode node11 = new TreeNode(); 12. node11.Text = "hopeoneone"; 13. TreeNode node2 = new TreeNode(); 14. node2.Text = "hopetwo"; 15. node1.Nodes.Add(node11);//在node1下面在添加一个结点。 16. node.Nodes.Add(node1);//node下的两个子节点。 17. node.Nodes.Add(node2); 18. 19. TreeNode t = new TreeNode("basil");//作为根节点。 20. treeView1.Nodes.Add(t); 21. TreeNode t1 = new TreeNode("basilone"); 22. t.Nodes.Add(t1);
Treeview 控件的简单应用： 在VB中Treeview 控件的添加： 通过VB菜单，[工程] -- [部件]，然后勾选Microsoft Windows Common Controls 6.0 （SP6），[确定]。 在工具箱里就有了Treeview 控件的图标。 Treeview 控件具有显示类似目录层次结构的格式，在具体应用中很有实际意义。下面简单介绍。 一．在Treeview 控件中添加1个新节点： 在Treeview 控件中添加1个新节点，是通过Treeview 控件的Nodes 集合的Add方法添加一个Node 对象来实现的。 使用方法： Dim nodX As Node Set nodX = Treeview1.Nodes.Add(relative, relationship, key, text, image, selectedimage) 其中参数说明： Relative：可选的。已存在的Node 对象的索引号或键值。新节点与已存在的节点间的关系，可在下一个参数relationship 中找到。可以这样理解relative的作用，是新节点的位置的参照对象。 Relationship：可选的。指定的Node 对象的相对位置，如设置值中所述。本参数是相对参数relative而言。 Key：可选的。唯一的字符串，可用于用Item 方法检索Node。 Text：必需的。在Node 中出现的字符串。 Image：可选的。在关联的ImageList 控件中的图像的索引。 Selectedimage：可选的。在关联的ImageList 控件中的图像的索引，在Node 被选中时显示。 以上参数image和selectedimage，是设置节点文字左边的图形，以后详细举例说明。 例1： Set nodX = TreeView1.Nodes.Add(, , "R", "Root") 这是缺省了relative, relationship, image, selectedimage参数的实例，而key值用“R”，text 值用“Root”。 通常，缺省了relative, relationship的节点，是第1层节点。本例，生成了一个第1层节点，显示的文字为“Root”。 例2： Set nodX = TreeView1.Nodes.Add("R", tvwChild, "C1", "Child 1") nodX.EnsureVisible '这个方法EnsureVisible，使得新添加的子节点后，展开多层显示。 本例：relative为“R”，relationship为tvwChild，key为“C1”，text为“Child 1” 其功能是：建立1个新节点；该新节点是节点key值为“R”的子节点（tvwChild），而该新节点的key值用“C1”，text值用“Child 1”。 我们现在要为TreeView1控件添加新节点，可能有三种情况； 1）添加1个新的第1层节点。 Set nodX = TreeView1.Nodes.Add(, , "R1", "Root1")
范道中学七年级数学导学提纲课题：幂的乘方 出卷人：施培新审核人：陈益锋 2012-2-22 姓名 _____ 课前参与 （一）预习内容：课本P43—44 （二）知识整理： 1.探索： （1）（2）是幂2的_____次方,其意义是_____个2的连乘积， 可写成：（2）=2×2=2= 2=2。 （2）（a）是幂a的_____次方,其意义是____个a的连乘积， 可写成：（a）=（）×（）×（）= a= a= a； 由此得：（a）是幂a的______次方,其意义是______个a的连乘积， 可写成：（a）=（）=a=a。 2.归纳：幂的乘方的法则：__________________________________________； 即写成公式： （a）=a（m、n为正整数）。 3.尝试练习： （1）（10）= （5）（－5）= （2）（10）= （6）（－5）= （3）（b）= (7 [（n－m）] 5 （4）（b）= (8 a·（a）2+ a·（a）3
4.推广：[（a m ）n ]p =____________ (m 、n 、p 为正整数。 5.幂的乘方法则的逆用为___________________________。 （三）思考： 通过预习，你认为本节内容主要研究了什么？你还有什么问题需和大家一起探讨？你有没有新的发现和大家一起分享！ 课中参与 例题1、计算：（1）（55）3 （2）（53）5 （3）(3x 5 （4）(35 x 例题2、计算：（1）[（a －b ）] （2）[（x －y ）] 例题3、计算：（1）－（y 4）3 （2）[（－y ）4]3 （3）（－y 4）3 例题4、计算：（1）（a ）·a （2）（b ）·（b ） （3）a ·（a ）－a ·（a ）2 拓展：1、（1）[（2）] （2）[（－3）] 2、已知3=2，3y =3，求（1）33x ，3 2y 的值。 （2）求3的值. 3、已知：3=a ，3=b ，用含a 、b 的代数式表示3 。 课后参与 课题：幂的乘方 姓名_____ 一、填空： （1）（7）5=_________； （2）[(－22]3=_________； （3） （a ） =________； （4）（－a 5）3=_________； （5）[（a －2）]=________； （6）[（x －y ）]=______；
Access 2003：Treeview控件（树型控件）快速入门(2010-06-01 14:26:08) 很多东西看起来很复杂，其实学起来还是蛮简单的。说这样的话不是“站着说话不腰疼”，而是切切实实的感受。很多时候我们会感到恍然大悟，之后便轻车熟路，信手拈来了，这就是前面所说的感觉，正所谓“山重水复疑无路，柳暗花明又一村”。 学这个Treeview控件也是一样。看起来它那么复杂，解释起来连篇累牍，但是我们需要的却往往只是其中一点，然后不断的重复使用这一点，仅仅这样，就能解决不少实际的问题。 使用Treeview的优点很多，比如具有无限扩展性，一个一个的分支，分支下面又可以增加次一级分支，每级分支又可以有很多并列的分支，这样就能满足多样的需求，另外，它还有很好的组织管理性，因为它具有明显的层级关系，很多人会用TreeView来做物料BOM表，可以说把这种特性发挥的淋漓尽致。 建立下面这样一个Treeview并不难，你只需要使用一句代码，多写几次就OK了。 不妨来看看代码，不过不用怕，记住，这里只有一句代码，其它的都在重复！ Dim ndeindex As Node Set ndeindex = TreeView0.Nodes.Add(, , "a", "基础资料", "k1")
Set ndeindex = TreeView0.Nodes.Add("a", tvwChild, "a1", "品号资料维护", "k1") Set ndeindex = TreeView0.Nodes.Add(, , "b", "工时资料", "k1") Set ndeindex = TreeView0.Nodes.Add("b", tvwChild, "b1", "观测资料查询", "k1") Set ndeindex = TreeView0.Nodes.Add("b", tvwChild, "b2", "工时查询（依品号）", "k1") Set ndeindex = TreeView0.Nodes.Add("b", tvwChild, "b3", "工时查询（依其它条件）", "k1") Set ndeindex = TreeView0.Nodes.Add(, , "c", "产能模式", "k1") Set ndeindex = TreeView0.Nodes.Add("c", tvwChild, "c1", "FCST产能计算", "k1") Set ndeindex = TreeView0.Nodes.Add("c", tvwChild, "c2", "产能试算", "k1") Set ndeindex = TreeView0.Nodes.Add(, , "d", "成本模式", "k1") 所以，不用太多解释，你应该明白这个函数的参数的意思了吧？第一个参数是指它从属的上级，如果它就是顶级，那就空着；第二个参数表示当前这个是前面的那个上级的下一级，这是系统规定的，照抄就行；第三个参数是当前级别的代号；第四个参数就是当前级别的显示文字，想看到什么就写什么；最后一个是指当前级别前面的图，这个在imagelist控件中，如果你要用，就加这个控件，不用也行，就把这个参数空着。（如果要用，需要现在imagelist中插入图像，然后再treeview中指定使用这个imagelist，k1是在插入图像时指定的图像代号。如下图所示）
电子系统设计实验报告 姓名 指导教师 专业班级 学院 提交日期2011年11月1日
目录 第一章设计题目 (1) 1.1 设计任务 (1) 1.2 设计要求 (1) 第二章原理分析及参数计算 (1) 2.1 总方案设计 (1) 2.1.1 方案框图 (1) 2.1.2 原理图设计 (1) 2.2 单元电路的设计及参数计算 (2) 2.2.1 二阶低通滤波器 (2) 2.2.2 二阶高通滤波器 (3) 2.3 元器件选择 (4) 第三章电路的组装与调试 (5) 3.1 MultiSim电路图 (5) 3.2 MultiSim仿真分析 (5) 3.1.1 四阶低通滤波器 (5) 3.1.2 四阶高通滤波器 (5) 3.1.3 总电路图 (6) 3.3 实际测试结果 (6) 第四章设计总结 (6) 附录………………………………………………………………………………… 附录Ⅰ元件清单………………………………………………………………… 附录Ⅱ Protel原理图…………………………………………………………… 附录Ⅲ PCB图（正面）………………………………………………………… 附录Ⅳ PCB图（反面）………………………………………………………… 参考文献…………………………………………………………………………
第一章 设计题目 1.1 设计任务 采用无限增益多重反馈滤波器，设计一四阶带通滤波器，通带增益01A =， 1L f kHz =，2H f kHz =，设计方案如图1.1所示。 图1.1 四阶带通滤波器方案图 1.2 设计要求 1.用Protel99 画出原理图，计算各元件参数，各元件参数选用标称值； 2.用Mutisum 对电路进行仿真，给出幅频特性的仿真结果； 3.在面包板上搭接实际电路，并测试滤波器的幅频特性； 4.撰写设计报告。 第二章 设计方案 2.1 方案设计 2.1.1方案框图(如图2.1.1) 图2.1.1 四阶带通滤波器总框图 2.1.2原理图设计 本原理图根据结构框图组成了4个二阶滤波器，上面两个分别为c f =2kHz ，Q=0.541，A=1的低通滤波器和c f =2kHz ，Q=1.306，A=1的低通滤波器；下面两个分别为c f =1kHz ，Q=0.541，A=1的高通滤波器和c f =1kHz ，Q=1.306，A=1的高通滤波器，其中P1、P2、P3作为接线座用来接线，原理图如图2.1.2,具体参数计算见2.2节。 V i V o 二阶低通滤波器 二阶低通滤波器 二阶高通滤波器 二阶高通滤波器
实验四微带线带通滤波器 设计 Prepared on 24 November 2020
实验四：基于ADS软件的平行耦合微带线带通滤波器的设计与仿真一、实验原理 滤波器是用来分离不同频率信号的一种器件，在微波电路系统中，滤波器的性能对电路的性能指标有很大的影响，微带电路具有体积小，重量轻、频带宽等诸多优点，在微波电路系统应用广泛，其中用微带做滤波器是其主要应用之一。平行耦合微带线带通滤波器在微波集成电路中是被广为应用的带通滤波器。 1、滤波器的介绍 滤波波器可以分为四种：低通滤波器和高通滤波器、带通滤波器和带阻滤波器。射频滤波器又可以分为以下波导滤波器、同轴线滤波器、带状线滤波器、微带滤波器。 滤波的性能指标： 频率范围：滤波器通过或截断信号的频率界限 通带衰减：滤波器残存的反射以及滤波器元件的损耗引起 阻带衰减：取通带外与截止频率为一定比值的某频率的衰减值 寄生通带：有分布参数的频率周期性引起，在通带外又产生新的通带 2、平行耦合微带线滤波器的理论 当频率达到或接近GHz时，滤波器通常由分布参数元件构成，平行耦合微带传输线由两个无屏蔽的平行微带传输线紧靠在一起构成，由于两个传输线之间电磁场的相互作用，在两个传输线之间会有功率耦合，这种传输线也因此称为耦合传输线。 平行耦合微带线可以构成带通滤波器，这种滤波器是由四分之一波长耦合线段构成，她是一种常用的分布参数带通滤波器。 当两个无屏蔽的传输线紧靠一起时，由于传输线之间电磁场的相互作用，在传输线之间会有功率耦合，这种传输线称之为耦合传输线。根据传输线理论，每条单独的微带线都等价为小段串联电感和小段并联电容。每条微带线的特性阻抗为Z0，相互耦合的部分长度为L，微带线的宽度为W，微带线之间的距离为S，偶模特性阻抗为Z e，奇模特性阻抗为Z0。单个微带线单元虽然具有滤波特性，但其不能提供陡峭的通带到阻带的过渡。 如果将多个单元级联，级联后的网络可以具有良好的滤波特性。 二、耦合微带线滤波器的设计的流程
delphi中Treeview的使用介绍 每一个节点下子节点形成这一节点的Items属性，当前节点有一个唯一的Index(TreeNode的Index属性)，用于说明子节点在Items 中的位置，每一个节点下的子节点是顺序编号的，第一个是0，第二个是1，依次类推。用IndexOf方法获得子节点的顺序，绝对顺序(AbsoluteIndex)则是指从Treeview第一个项开始的顺序值，第一个是0，如此推下去。Item属性则根据Index的值返回当前节点的第Index个子节点。Count则表明属于此项的所有子节点的数量。用MoveTo方法将Item由一个位置移到另一个位置。 Expanded属性表明是否所有的子项都全部展开(包括子项的子项)，为True表示全部展开。IsVisible属性表明一个项是否在树中能被看到，如果树全部展开那么这个Item是肯定可以被看到。HasChildren属性表明一个项是否有子项。 GetFirstChild, GetLastChild, GetPrevChild, and GetNextChild分别返回当前项子项的第一个、最后一个和前一个、后一个项。GetNextSibling and GetPrevSibling则返回在同一Level下的下一个和上一个项。GetNextVisible and GetPrevVisible则返回能看得到的下一个和上一个项。如果一个节点有Parent，则HasAsParent方法返回True. Parent为当前项的父项。Focused属性确定焦点是否落在此节点上，被Focus时会一个标准的方框围住。很显然，只有一个节点会被聚焦。Selected属性表明一个节点是否被选中，同样只有一个节点会被选中。DropTarget属性表明节点在拖动操作中是源还是目标。
DELPHI中利用TreeView控件建立目录树 p>Rainbow的话：关于TreeView的使用，还可以参看：联合使用TreeView组件 TreeView是一个显示树型结构的控件，通过它能够方便地管理和显示具有层次结构的信息，是Windows应用程序的基本控件之一。DELPHI虽然具有比较强大的文件管理功能，提供了多个用于文件管理的标准控件，如DriveComboBox、DirectoryListBox、FileListBox等，通过设置它们的属性，使其建立起联系，甚至不用编写一行程序，我们就可以实现在不同的目录之间进行切换，然而这样的目录切换只适用于进行文件的查找定位，而不能方便地进行目录的浏览，例如我们要从c:\windows目录转到c:\program files目录，就必须返回到根目录才能进行切换，而不能象Windows资源管理器那样任意地在不同的目录之间进行浏览与切换。 要实现在不同目录之间任意切换和浏览，还是需要使用TreeView控件，以下程序就利用DELPHI的TreeView控件来建立目录树。 在该程序中采用的各部件以及界面设计如下图所示： 各部件的主要属性设置如下： 该程序利用DriveCommboBox控件来获得系统具有的驱动器，并以此作为目录树的最上层，利用FileListBox控件，通过设置其Filetype属性为fdDirectory，可以获得所需的子目录，在TreeView控件的OnExpanding事件中将得到的子目录加到该控件的某一节点下。 整个程序的源代码如下： unit main;
interface uses Windows,Messages,SysUtils,Classes,Graphics,Controls,Forms,Dialogs,StdCtrls,FileCtrl,ComCtrls,ImgList; type TForm1=class(TForm) DirTreeView:TTreeView; FileListBox1:TFileListBox; DriveComboBox1:TDriveComboBox; ImageList1:TImageList; procedure FormCreate(Sender:TObject); procedure DirTreeViewExpanding(Sender:TObject;Node:TTreeNode;var AllowExpansion:Boolean); private {Private declarations} public {Public declarations} end; var Form1:TForm1; implementation {$R*.DFM} procedure TForm1.FormCreate(Sender:TObject); var FirstNode,DirNode:TTreeNode; ItemCount,Index:integer; Itemstr:string; begin ItemCount:=DriveComboBox1.Items.Count;//所有驱动器的个数 FirstNode:=DirTreeView.Items.GetFirstNode; for index:=0to ItemCount-1do begin ItemStr:=DriveComboBox1.Items[index]; ItemStr:=copy(ItemStr,1,pos(:,ItemStr));//获得驱动器的名称（比如C/D） DirNode:=DirTreeView.Items.AddChild(FirstNode,ItemStr); DirNode.HasChildren:=true; DirNode.ImageIndex:=0; DirNode.SelectedIndex:=1; end; end; //响应扩展事件 procedure TForm1.DirTreeViewExpanding(Sender:TObject;Node:TTreeNode;Var AllowExpansion:Boolean); var