The Opsis Project Materialized Views for Data Warehouses and the Web

The Opsis Project:Materialized Views for

Data Warehouses and the Web

Nick Roussopoulos,Yannis Kotidis,Alexandros Labrinidis,and Yannis Sismanis

Department of Computer Science

University of Maryland

College Park,MD20742,USA

nick,labrinid,isis@https://www.sodocs.net/doc/5f4541654.html,

AT&T Labs Research

180Park Ave,P.O.Box971,

Florham Park,NJ07932,USA

kotidis@https://www.sodocs.net/doc/5f4541654.html,

Abstract.The real world we live in is mostly perceived through an incredibly large col-

lection of views generated by humans,machines,and other systems.This is the view re-

ality.The Opsis project concentrates its efforts in dealing with the multifaceted form and

complexity of data views including data projection views,aggregateviews,summary views

(synopses),point of view views,and?nally web views.In particular,Opsis deals with

the generation,the storage organization(Cubetrees),the ef?cient run-time management

(Dynamat)of materialized views for Data Warehouse systems,and for web servers with

dynamic content(WebViews).

1Introduction

Most of the data stored and used today is in the form of materialized views,generated from sev-eral possibly distributed and loosely coupled source databases.These views are sorted and or-ganized appropriately in order to rapidly answer various types of queries.The relational model is typically used to de?ne each view and each de?nition serves a dual purpose,?rst as a speci-?cation technique and second as an execution plan for the derivation of the view data.

Materialized views are approximately10years younger than the relational model.Early papers that foresaw their importance include[26,25,7,29,34,31,30].During this period,ma-terialized views were considered by top relationalists as the“Pandora’s box”.It took another 6-7years before it was realized how useful and versatile they were.Then,a?urry of papers rehashed the earlier results and almost brought the research on materialized views to extinc-tion.But,materialized views were too important and research continues as of today.Relational views have several forms:

–pure program:an unmaterialized view is a program speci?cation,“the intention”,that gen-erates data.Query modi?cation[32]and compiled queries[2]were the?rst techniques ex-ploiting views–their basic difference is that the?rst is used as a macro that does not get optimized until run-time,while the second stores optimized execution plans.Each time

the view program is invoked,it generates(materializes)the data at a cost that is roughly the same for each invocation.

–derived data:a materialized view is“the extension”of the pure program form and has the characteristics of data like any other relational data.Thus,it can be further queried to build views-on-views or collectively grouped to build super-views.The derivation operations are attached to materialized views.These procedural attachments along with some“delta”re-lational algebra are used to perform incremental updates on the extension.

–pure data:when materialized views are converted to snapshots,the derivation procedure is detached and the views become pure data that is not maintainable(pure data is at the opposite end of the spectrum from pure program).

–pure index:view indexes[26]and ViewCaches[27]illustrate this?avor of views.Their extension has only pointers to the underlying data which are dereferenced when the values are needed.Like all indexing schemes,the importance of indexes lies in their organization, which facilitates easy manipulation of pointers and ef?cient single-pass dereferencing,and thus avoids thrashing.

–hybrid data&index:a partially materialized view[3]stores some attributes as data while the rest are referenced through pointers.This form combines data and indexes.B-trees,Join indexes[35],star-indexes and most of the other indexing schemes belong to this category, with appropriate schema mapping for translating pointers to record?eld values.

–OLAP aggregate/indexing:a data cube[10]is a set of materialized or indexed views[12, 23,18].They correspond to projections of the multi-dimensional space data to lesser di-mensionality subspaces and store aggregate values in it.In this form,the data values are ag-gregated from a collection of underlying relation values.Summary tables and Star Schemas belong in this form as well.

–WebViews:HTML fragments or entire web pages that are automatically created from base data,typically stored in a DBMS[20].Having a WebView materialized can potentially give signi?cantly lower query response times,compared computing it on the?y.However,it may also lead to performance degradation,if the update workload is too high.

Each of these forms is used by some component of a relational system.Having a uni?ed view of all forms of relational views is important in recognizing commonalities,re-using im-plementation techniques,and discovering potential uses not yet exploited.The Opsis project has focused on developing storage and update techniques for all forms of materialized views.We have been particularly careful with the ef?cient implementation and scalability of these meth-ods.We have architected,designed,implemented,and tested giant-scale materialized view en-gines for the demands of todays abundance of connectivity and data collection.

This paper is organized as follows.In the next section we describe the Cubetree Data Model, a storage abstraction for the data cube,and also present a compact representation for it using packed R-trees[24].In Section3,we present our algorithm for bulk incremental updates of the data cube.Section4has a brief outline of DynaMat,a view management system that materi-alizes results from incoming aggregate queries as views and exploits them for future reuse.In Section5we explore the materialization policies for WebViews and present results from ex-periments on an industrial-strength prototype.Section6discusses the Quality of Service and Quality of Data considerations for WebViews.Finally,we conclude in Section7.

2A Storage Abstraction for OLAP Aggregate Views

Consider the relation where,,and are the grouping attributes that we

would like to compute the cube for the measure attribute.We represent the grouping at-

tributes,,and on the three axes of x x and then map each tuple of

using the values,,for coordinates and the value as the content of the data point.

We now project all the data points on all subspaces of AxBxC and aggregate their content.We assume that each domain of R has been extended to include a special value(zero in this exam-

ple)on which we do the projections.A projection on a subspace with dimension,

where is the number of grouping attributes,represents the group by of all those attributes that

correspond to.The aggregate values of are stored in the intersection points between

and the orthogonal-dimensional hyper-planes that correspond to the remaining

dimensions not included in.For example,the projection planes,,...parallel to plane x shown in Figure1,correspond to group by and their aggregated values are stored in the content of their intersection point with axis.The origin is used to store the

(super)-aggregate value obtained by no grouping at all.We call this the Cubetree Data Model

(CDM).

Fig.1.group by

projections Fig.2.Querying the Cube

In CDM,we map cube and relational queries into multi-dimensional range queries.For ex-ample,a query to?nd all the group by values for between3and6would be formulated as a range query shown by the bold-dashed line in Figure2.If now we would like to?nd out the percent contribution(multidimensional ratio)of to these group by values,we obtain the intersection points of line with planes,,etc. and the content of them is divided by the corresponding aggregates on.

Clearly,different combinations of relational,1-dimensional or multi-dimensional storage structures can be used to realize the CDM.For example,the whole CDM can be realized by just a conventional relational storage[10]with no indexing capability for the cube.Another possibility,would be to realize CDM by an R-tree[14],or a combination of relational structures, R-trees and B-trees[5].Since most of the indexing techniques are hierarchical,without loss of generality,we assume that the CDM is a tree-like(forest-like)structure that we refer to as the Cubetree of.

3Bulk Incremental Updates

Random record-at-a-time insertions are not only very slow because of the continuous reorgani-zation of the space,but also destroy data clustering in all multidimensional indexing schemes. Packed R-trees,introduced in[24],avoid these problems by?rst sorting the objects in some desirable order and then bulk loading the R-tree from the sorted?le and packing the nodes to capacity.This sort-pack method achieves excellent clustering and signi?cantly reduces the overlap and dead space(i.e.space that contains no data points).

The proposed bulk incremental update computation is split into a sort phase where an up-date increment of relation is sorted,and a merge-pack phase where the old Cubetree is packed together with the updates:Merge-Pack sort Sorting could be the dominant cost factor in the above incremental computation,but it can be parallelized and/or con?ned to a quantity that can be controlled by appropriate schedules for re-freshing the cube.Note that contains any combination of relation insertions,deletions,and updates.For aggregate functions that are Self Maintainable[21]like count()and sum(),they are all equivalent because they all correspond to a write of all projection points with their content adjusted by appropriate arithmetic expressions.

We assume that a tuple in has the following structure:,where ,denotes the value on dimension,=and is the measure attribute.The gener-alization with more measure attributes is straightforward.During the sorting phase,we read and create a sorted run for each group by in the Cube.The format of a tuple in is: ,where,denotes the values of each di-

mension in that group by.For a speci?ed point in this-dimensional space

hold the aggregated values.In order to be able to merge with the existing aggregates that are stored within the Cubetrees,the data within each update increment are being sorted in the same order as the data within the Cubetrees.For example,if all points of group by ABC are stored in the order,then the same sort order is being used for the new projections from the deltas.

During the second merge-pack phase the old Cubetrees are packed together with the up-dates.For each Cubetree,all increments that are stored in that Cubetree are opened and merge-packed with to create a new instance of the index.

3.1Creation/Maintenance Measurements for a Grocery Demo Dataset

We used a synthetically generated grocery demo dataset that models supermarket transactions. The data warehouse is organized according to the star schema[17]organization.There is a sin-gle fact table sales that includes12dimension attributes and two real(4-byte)measure at-tributes,namely revenue and cost.We pre-computed several aggregate views and stored them within Cubetrees.These views aggregate data over attributes chosen from the dataset and compute the sum()aggregate for both measures.1

We ran the experiment on a360MHz Ultra SPARC60with two SCSI18GB Seagate Chee-tah hard drives.For sorting the data,we used a simple quick-sort utility.There are numerous optimizationsthat we could have exploited for speeding up this phase,see[1,22].Table1shows

the time for the initial load of the Cubetrees and the time for each bulk-incremental update with a year’s and?ve month’s worth of data.The corresponding sizes for each update increment are also given.

Transactions Total Records Pack Time Packing rate 1/1/90-12/31/97127,702,7085m51s29.95GB/h 1/1/98-12/31/98143,216,7897m:54s25.95GB/h 1/1/99-5/31/99160,587,1438m53s25.88GB/h

Table1.Initial Bulk-load and Bulk-incremental Updates

For sorting and packing we utilized both Seagate Cheetah disks i.e.reading the input from one disk and packing the Cubetrees in the other.The size of the views for the initial creation was2.92GB and the packing phase was completed in less than6minutes.This corresponds to a packing rate(speed)of29.95GB/h or8.52MB/sec,roughly68%of the raw serial disk write rate. The remaining bandwidth is lost due to the necessary processing of the input.The second and third lines of the table show the performance during bulk-incremental updates.The effective disk packing rate that we got for updates was slightly slower,at about26GB/h.This is because we only used two disks,storing the input data(updates)in the?rst and the Cubetrees in the second.Therefore,during updates both the old and the new-version of the Cubetrees were on the same disk sharing its I/O bandwidth.

4Dynamic Management of Aggregate Views

Disk space and creation/maintenance overhead will not allow us to materialize all interesting group-bys of the data cube.The view selection problem[26,15,4,33,16]consists of?nding those group-bys that minimize query response time under a resource constraint(typically disk space)and store them as materialized views.

This static selection of views however,contradicts the dynamic nature of decision support analysis.Especially for ad-hoc queries where the expert user is looking for interesting trends in the dataset,the query pattern is dif?cult to predict.Furthermore,as query patterns and data trends change overtime and as the data warehouse is evolving with respect to new business re-quirements that continuously emerge,even the most?ne-tuned selection of views that we might have obtained at some point,will very quickly become outdated.In addition,the maintenance window,the disk space restrictions and other important functional parameters of the system also change.For example,an unexpected large volume of daily updates will throw the selected set of views as not update-able unless some of these views are discarded.

Another inherit drawback of a static view selection scheme is that the system has no way of tuning a wrong selection by re-using results of queries that couldn’t be answered by the ma-terialized set.Notice that although OLAP queries take an enormous amount of disk I/O and CPU processing time to be completed,their output is often quite small as they summarize the underlying data.Moreover,during roll up operations[10]the data is examined at a progres-sively coarser granularity and future queries are likely to be computable from previous results without accessing the base tables at all.

In[19]we introduced DynaMat,a dynamic view management system for data warehouses. Our work has been motivated by earlier research on caching and resusing query results in rela-tional database systems[31,9].DynaMat manages a dedicated disk space that we call the View

Pool,in which previously computed aggregates are stored.There are two distinct modes of op-eration.The?rst is the“on-line”mode during which user queries are allowed.DynaMat deter-mines whether or not aggregates(views)stored in the View Pool can be used in a cost-effective manner to answer a new query,in comparison to running the same query against the detailed records in the data warehouse.This is achieved by probing the query-optimizer and getting an estimate of the execution cost of the query at the data warehouse.Whenever a new query result is computed,DynaMat uses an admission/replacement strategy that exploits spatio-temporal locality in the user access pattern,but also takes into account the computational dependencies of the stored query results.

Periodically,updates received from the data sources get shipped to the data warehouse and the View Pool gets refreshed.During updates,DynaMat switches to an“off-line”mode during which queries are not permitted.The maximum length of the update phase is speci?ed by the administrator.Different update policies are implemented,depending on the types of updates,the properties of the data sources and the aggregate functions that are computed by the query results. From DynaMat’s point of view the goal is to select and update the most useful fragments within the update time constraint.Notice that this is not equivalent to updating as many fragments as possible,although often both yield similar results.

5WebView Materialization

WebViews are HTML fragments that are automatically created from based data,which are typ-ically stored in a DBMS.For example,a search at an online bookstore for books by a particular author returns a WebView that is generated dynamically;a query on a cinema server generates a WebView that lists the current playing times for a particular movie;a request for the current sports scores at a newspaper site returns a WebView which is generated on the?y.Except for generating web pages as a result of a speci?c query,WebViews can also be used to produce multiple versions(views)of the same data(for example,translating the contents of a web page in multiple languages),and to support multiple web devices,especially browsers with limited display or bandwidth capabilities,such as cellular phones or networked PDAs.

Although there are a few web servers that support arbitrary queries on their base data,most web applications“publish”a relatively small set of prede?ned or parameterized WebViews, which are to be generated automatically through DBMS queries.A weather web server,for ex-ample,would most probably report current weather information and forecast for an area based on a ZIP code,or a city/state combination.Given that weather web pages can be very popu-lar and that the update rate for weather information is not high,materializing such WebViews would most likely improve performance.

Personalized WebViews[6]can also be considered for materialization,if?rst they are de-composed into a hierarchy of WebViews.Take for example a personalized newspaper.It can have a selection of news categories(only metro,international news),a localized weather fore-cast(for Washington,DC)and a horoscope page(for Scorpio).Although this particular com-bination might be unique or unpopular,if we decompose the page into four WebViews,one for metro news,one for international news,one for the weather and one for the horoscope,then these WebViews can be accessed frequently enough to merit materialization.

5.1WebView Materialization Policies

We explore three materialization policies:virtual,materialized inside the DBMS and materi-

alized at the web server.In the virtual policy,everything is computed on the?y.To produce

a WebView we must query the DBMS and format the results in HTML.Since no views are cached,we only need to update the base tables,whenever there is an update.

In the materialized inside the DBMS policy,we save the results of the SQL query that is

used to generate the WebView.To produce the WebView,we must access the stored results and

format them in HTML.The main difference of WebView materialization from web caching is

that,in the materialization case,the stored query results need to be kept up to date all the time.

This leads to an immediate refresh of the materialized views inside the DBMS with every update

to the base tables they are derived from.

Finally,in the materialized at the web server policy,in order to satisfy user requests we simply have to read the WebView from the disk,where a fresh version is expected to be stored.

This means that on every update to one of the base tables that produce the WebView,we have

to refresh the WebView(or recompute it,if it cannot be incrementally refreshed)and save it as

a?le for the web server to read.

5.2The selection problem

The choice of materialization policy for each WebView has a big impact on the overall perfor-

mance.For example,a WebView that is costly to compute and has very few updates,should be

materialized to speed up access requests.On the other hand,a WebView that can be computed fast and has much more updates than accesses,should not be materialized,since materialization

would mean more work than necessary.We de?ne the WebView selection problem as follows: For every WebView at the server,select the materializationstrategy(virtual,material-

ized inside the DBMS,materialized at the web server),which minimizes the average

query response time on the clients.

We assume that there is no storage constraint,since storage means disk space(not main mem-

ory),and also WebViews are expected to be relatively small.The decision whether to mate-

rialize a WebView or not,is similar to the problem of selecting which views to materialize in

a data warehouse[11,13,28],known as the view selection problem.There are two crucial dif-

ferences.First of all,the multi-tiered architecture of typical database-backed web servers raises the question of where to materialize a WebView.Secondly,updates are performed online at web servers,as opposed to data warehouses which are usually off-line during updates.

5.3Experiments

In[20]we considered the full spectrum of materialization choices for WebViews in a database-backed web server.We compared them analytically using a detailed cost model that accounts for both the inherent parallelism in multitasking systems and also for the fact that updates on the base data are to be done concurrently with the accesses.We have implemented all?avors of WebView materialization and run extensive experiments on an industrial strength prototype, based on the Apache web server and Informix,running on a SUN UltraSparc-5with320MB of memory and a3.6GB hard disk.We used a pool of22SUN Ultra-1workstations as clients. Due to space constraints we only present two of our experiments.

Fig.3.Scaling up the access rate mat-db

0510152025

In the?rst set of experiments,we varied the

and measured the average query response time

cies:virtual(virt),materialized inside the

(mat-web).A load of50requests/sec corresponds to a rather“heavy”web server load of4.3 million hits per day for dynamically generated pages.The incoming update rate was5up-dates/sec for all experiments.

In Figure3we report the average query response time per WebView as they were measured at the web server.We immediately notice that the mat-web policy has average query response times that are consistently at least an order of magnitude(10-230times)less than those of the virt or mat-db policies.Under the mat-web policy,servicing a request involves simply read-ing a?le from disk,whereas,under the virt and mat-db policies,the system needs to compute a query at the DBMS for every request.Furthermore,since the web processes are“lighter”than the DBMS processes,the mat-web policy scales better than the other two.

In the second set of experiments,we varied the incoming update rate from0to25updates/sec, while the access request rate was set at25accesses/sec.In Figure4we plot the average query response times for this experiment under the three materialization policies.Our?rst observa-tion is that the average query response time remains practically unchanged for the mat-web policy despite the updates,because the updates are performed in the background.The second observation is that the virt policy is performing56%-93%better than the mat-db policy in the presence of updates.This is explained by the fact that updates under the mat-db policy lead to extra work at the DBMS in order for the materialized views to be kept up to date.

6Measuring the Quality of Web Servers

Caching of static web pages[8]is known to improve the Quality of Service(QoS)for user re-quests,since it improves the average query response time.For dynamic content however,web caching does not provide any freshness guarantees on the cached data.Servicing user requests fast is of paramount importance only if the data is fresh and correct,otherwise it may be more harmful than slow or even no data service.In general,when measuring the Quality of a system that uses materialized views,we need to evaluate both QoS and the Quality of Data(QoD),or how“fresh”the served data are.Web caching improves QoS dramatically,but completely ig-nores QoD of the cached data.On the other hand,when QoD is very important,web servers rely

on computing frequently changing web data on-demand.This achieves near-perfect QoD,but seriously impedes performance or leads to server melt-downs.WebView Materialization[20, 36]aims at bridging this gap,since it prolongs the QoS bene?ts of caching using amortization and incremental update algorithms on the cached data.This improves the QoD at a small degra-dation in QoS.In our work we try to provide the best trade-off between QoS and QoD based on the user/application requirements and the incoming workload.

7Conclusions

In this paper we concentrated on the most important feature of the relational database model: materialized views.We focused on their usage on data warehousing and Web servers,with em-phasis on updateability,performance,and scalability.

Speci?cally,we presented the Cubetree Data Model,a storage abstraction for the data cube. Cubetrees maintain the view records internally sorted(using packed R-trees)and allow bulk incremental updates through an ef?cient merge-packing algorithm.We brie?y described Dyna-Mat,a dynamic view management system that manages collections of materialized aggregate views based on user workload and available system resources(disk space,update cost).Finally, we explored the materialization policies for WebViews,presented experimental results from an industrial-strength prototype and discussed the Quality of Service and Quality of Data consid-erations for WebViews.

References

1.S.Agrawal,R.Agrawal,P.Deshpande,A.Gupta,J.Naughton,R.Ramakrishnan,and S.Sarawagi.

On the Computation of Multidimensional Aggregates.In Proc.of22nd VLDB conference,pages 506–521,Bombay,India,August1996.

2.M.M.Astrahan et al.System R:Relational Approach to Database Management.ACM Transactions

on Database Systems,1(2):97–137,June1976.

https://www.sodocs.net/doc/5f4541654.html,rs Baekgraard and Nick Roussopoulos.“Ef?cient Refreshment of Data Warehouse Views”.Tech-

nical Report CS-TR-3642,Dept.of Computer Science,Univ of Maryland,May1996.

4.E.Baralis,S.Paraboschi,and E.Teniente.Materialized View Selection in a Multidimensional Data-

base.In Proc.of the23th VLDB Conference,Athens,Greece,August1997.

5.R.Bayer and https://www.sodocs.net/doc/5f4541654.html,anization and Maintenance of Large Ordered Indexes.Acta Infor-

matica,1(3):173–189,1972.

6.Phil Bernstein et al.“The Asilomar Report on Database Research”.SIGMOD Record,Dec.1998.

7.Jos′e A.Blakeley,Per˙Ake Larson,and Frank Wm.Tompa.“Ef?ciently Updating Materialized

Views”.In Proc.of the ACM SIGMOD Conference,pages61–71,Washington,DC,May1986.

8.Lee Breslau,Pei Cao,Li Fan,Graham Phillips,and Scott Shenker.“Web Caching and Zipf-like Dis-

tributions:Evidence and Implications”.In Proc.of INFOCOM’99,New York,USA,March1999.

9.A.Delis and N.Roussopoulos.Performance and Scalability of Client-Server Database Architectures.

In Proc.of the18th VLDB,pages610–623,Vancouver,Canada,1992.

10.J.Gray,A.Bosworth,https://www.sodocs.net/doc/5f4541654.html,yman,and H.Pirahesh.Data Cube:A Relational Aggregation Operator

Generalizing Group-By,Cross-Tab,and Sub-Totals.In Proc.of the12th ICDE,pages152–159,New Orleans,February1996.IEEE.

11.Ashish Gupta and Inderpal Singh Mumick.“Maintenance of Materialized Views:Problems,Tech-

niques,and Applications”.Data Engineering Bulletin,18(2):3–18,June1995.

12.H.Gupta,V.Harinarayan,A.Rajaraman,and J.Ullman.Index Selection for OLAP.In Proceedings

of ICDE,pages208–219,Burmingham,UK,April1997.

13.Himanshu Gupta.“Selection of Views to Materialize in a Data Warehouse”.In Proc.of the6th

International Conference on Database Theory(ICDT’97),Delphi,Greece,January1997.

14.A.Guttman.R-Trees:A Dynamic Index Structure for Spatial Searching.In Proc.of the ACM SIG-

MOD International Conference on Management of Data,Boston,MA,June1984.

15.V.Harinarayan,A.Rajaraman,and J.Ullman.Implementing Data Cubes Ef?ciently.In Proc.of ACM

SIGMOD,pages205–216,Montreal,Canada,June1996.

16.H.J.Karloff and M.Mihail.On the Complexity of the View-Selection Problem.In Proceedings

of the18th ACM SIGACT-SIGMOD-SIGART Symposium on Principles of Database Systems,pages 167–173,Philadelphia,Pennsylvania,May1999.

17.R.Kimball.The Data Warehouse Toolkit.John Wiley&Sons,1996.

18.Y.Kotidis and N.Roussopoulos.An Alternative Storage Organization for ROLAP Aggregate Views

Based on Cubetrees.In Proceedingsof the ACM SIGMOD International Conferenceon Management of Data,pages249–258,Seattle,Washington,June1998.

19.Yannis Kotidis and Nick Roussopoulos.DynaMat:A Dynamic View Management System for Data

Warehouses.In Proceedings of the ACM SIGMOD International Conference on Management of Data,pages371–382,Philadelphia,Pennsylvania,June1999.

20.Alexandros Labrinidis and Nick Roussopoulos.“WebView Materialization”.In Proc.of the ACM

SIGMOD Conference,Dallas,Texas,USA,May2000.

21.I.S.Mumick,D.Quass,and B.S.Mumick.Maintenance of Data Cubes and Summary Tables in a

Warehouse.In Proc.of the ACM SIGMOD Conference,Tucson,Arizona,May1997.

22.K.A.Ross and D.Srivastava.Fast Computation of Sparse Datacubes.In Proceedings of the23th

VLDB Conference,pages116–125,Athens,Greece,Augoust1997.

23.N.Roussopoulos,Y.Kotidis,and M.Roussopoulos.Cubetree:Organization of and Bulk Incremen-

tal Updates on the Data Cube.In Proceedings of the ACM SIGMOD International Conference on Management of Data,pages89–99,Tucson,Arizona,May1997.

24.N.Roussopoulos and D.Leifker.Direct Spatial Search on Pictorial Databases Using Packed R-trees.

In Procs.of1985ACM SIGMOD,pages17–31,Austin,1985.

25.Nick Roussopoulos.“The Logical Access Path Schema of a Database”.IEEE Transactions on Soft-

ware Engineering,8(6):563–573,November1982.

26.Nick Roussopoulos.“View Indexing in Relational Databases”.ACM Transactions on Database Sys-

tems,7(2):258–290,June1982.

27.Nick Roussopoulos.“An Incremental Access Method for ViewCache:Concept,Algorithms,and Cost

Analysis”.ACM Transactions on Database Systems,16(3),September1991.

28.Nick Roussopoulos.“Materialized Views and Data Warehouses”.SIGMOD Record,March1998.

29.Nick Roussopoulos and Hyunchul Kang.“Principles and Techniques in the Design of ADMS”.

Computer,pages19–25,December1986.

30.A.Segev and J.Park.Maintaining Materialized Views in Distributed Databases.In Proceedings of

the5th International Conference on Data Engineering,Los Angeles,CA,1989.

31.Timos Sellis.“Intelligent caching and indexing techniques for relational database systems”.Infor-

mation Systems,13(2),1988.

32.Michael Stonebraker.“Implementation of Integrity Constraints and Views by Query Modi?cation”.

In Proc.of the ACM SIGMOD Conference,San Jose,California,May1975.

33.D.Theodoratos and T.Sellis.Data Warehouse Con?guration.In Proc.of the23th International

Conference on VLDB,pages126–135,Athens,Greece,August1997.

34.F.Tompa and J.Blakeley.Maintaining Materialized Views Without Accessing Base https://www.sodocs.net/doc/5f4541654.html,rma-

tion Systems,13(4):393–406,1988.

35.Patrick Valduriez.“Join Indices”.ACM Transactions on Database Systems,12(2),June1987.

36.Khaled Yagoub,Daniela Florescu,Valerie Issarny,and Patrick Valduriez.“Caching Strategies for

Data-Intensive Web Sites”.In Proc.of the26th VLDB Conference,Cairo,Egypt,Sep2000.

牛津译林版高中英语必修一模块一

高中英语学习材料 madeofjingetieji 盐城市龙冈中学牛津高中英语模块一测试题 命题人：程永华 第一卷(选择题，共75分) I.听力（共两节，满分20分） 第一节（共5小题；满分5分） 听下面5段对话。每段对话后有一个小题，从题中所给的A、B、C三个选项中选出最佳选项，并标在试卷的相应位置。听完每段对话后，你都有10秒钟的时间来回答有关小题和阅读下一小题。每段对话仅读一遍。 1.How much is the skirt? A.$19.15. B.$9.15. C.$9.50 2.What are the two speakers doing? A.They are having a rest. B.They want to have dinner. C.They are hiking on a hill. 3.What does the man mean? A.He doesn't like Chinese food. B.He doesn't want to eat out. C.He wants to pay for the meal. 4.What did the man do yesterday evening? A.He had dinner with his friend. B.He visited a friend. C.He saw a film. 5.When does Mr Brown have to be at the airport? A.At 8:30. B.At 9:30. C.At 10:30. 第二节（共15小题，满分15分） 听下面5段对话或独白。每段对话或独白后有几个小题，从题中所给的A、B、C三个选项中选出最佳选项，并标在试卷的相应位置。听每段对话或独白前，你将有时间阅读各个小题，每小题5秒钟；听完后，各小题将给出5秒钟的作答时间。每段对话或独白读两遍。 听第6段材料，回答第6至8题。 6.What kind of color TV does the woman want to buy? A.A small one. B.A big one. C.A cheap one. 7.What's the man's opinion about buying a TV? A.The woman should buy a small one. B.The woman should buy a big one. C.The woman should buy a cheap one. 8.Why does the man give the woman such a suggestion? A.Because the big one is too expensive. B.Because she will have to change it in a few years. C.Because her sitting room isn't very big. 听第7段材料，回答第9至11题。 9.Where is the boy going to spend his holiday? A.In the city. B.In his hometown. C.Abroad. 10.What will he do there? A.Do some farm work. B.Study at home. C.Do some housework. 11.Which of the following is TRUE? A.The girl knows how to farm. B.The boy's grandparents don't live in the city. C.The boy's school is in the country. 听第8段材料，回答第12至14题。 12.What are the man and the woman looking for? A.A comfortable hotel. B.A modern hotel. C.A cheap hotel. 13.Where could the man and the woman find the hotel they need? A.Next to the bank. B.A two minutes' drive to the west. C.On the left side of the street.

(完整版)新课标人教版高中英语选修6单词-中文

选修六 Unit 1 1.adj. 现实主义的；逼真的；现实的 2.adj. 抽象的；深奥的n. 摘要 3.n. 雕塑 4.n. 雕刻家；雕塑家 5.n. 美术陈列室；画廊 6.n. 信任；信心；信念 7.adv. 忠实地 8.adv. 所以；因而 9.n. 目标；目的vi. & vt. 瞄准；（向某方向）努力 10.adj. 常规的；传统的；因循守旧的 11.adj. 典型的；有代表性的 12.adj. 明显的；明白的 13.n. 新生；复兴；复活 14.vt. 采用；采纳；收养 15.adj. 人道主义的 16.vt. 拥有；具有；支配 17.n. （尤作复数）所有；财产 18.adj. 卓越的；杰出的；极好的 19.n. 透视画法；透视图；观点 20.n. 技术；方法；技能 21.n. 巧合（的事）；（事情、口味、故事等）相合 22.巧合地 23.n. 杰作；名著 24.n. 印象主义；印象派 25.adj. 印象派的 n.印象派艺术家 26.adj. 后印象派的n. 后印象派艺术家 27.大量28.n. 阴影；影子 29.adj. 荒谬的；可笑的 30.adj. 争论的；争议的 31.n. 努力；尝试；企图vt. 尝试；企图 32.（可是）另一方面 33.vt. 预言；预告；预测 34.n. 风景；景色 35.adj. 确切的；特定的 36.n. 画像；身材；数字 37.n. 黏土 38.n. 评论家；批评者 39.n. 青铜；青铜色；青铜制艺术品 40.n. 大理石 41.vt. 雕刻；刻记 42.adj. 脆弱的；容易生病的；精致的 43.n. 帆布；画布 44.n. 咖啡馆；小餐馆 45.adj. 过敏性的；对……过敏的 46.adv. 有效地 47.n. 展览；陈列；展览会 48.adj.敢做敢为的;侵略的；好斗的 49.n. 学者 50.n. 肉；肌肉；肉体 51.活着的；本人 52.n. 几何学 53.n. 束；串 54.n. 林荫道；道路；大街 55.n. 喜爱；偏爱

project项目管理

实验五软件项目管理工具 [掌握与了解] Project Professional 2003 是一个企业项目管理程序，它与Project Server 2003 和Project Web Access 一起帮助您更有效地计划项目，更好地管理资源，并且在整个组织范围内建立可以重复的过程。 功能有效地管理项目：通过利用集中式项目模板储备库，从而利用最佳的惯例。利用分步式指南有效地创建、管理项目并为之提供资源。 建立规则和过程：锁定时间段并确定多个批准人，从而保持已完成工作的精确记录。 更有效地管理资源：通过集中式资源库，更好地掌握职员的技能水平和可用性。利用基于技能的资源分配工具为项目提供资源。 计划资源需求：通过计划和跟踪资源要求，确保有合适的人员参与未来的项目。 [课前准备] （1）项目进度计划 项目进度计划的主要工作见下图： （2）横道图（甘特进度图） 甘特图表/甘特进度图（Gantt chart）是在20世纪初由亨利甘特开发的。它基本上是一种线条图，横轴表示时间，纵轴表示要安排的活动，线条表示在整个期间上计划的和实际的活动完成情况。甘特图直观地表明任务计划在什么时候进行，以及实际进展与计划要求的对比。 横道图有以下特点： ?编制比较容易，绘图比较简单。 ?表达形象直观，排列整齐有序。 ?便于用叠加法进行资源需要量的统计。

（3）网络计划图 网络计划图是由一系列的箭线和节点所组成的网状图形，用来表示各个施工过程在各施工段上各项工作的起止时间和先后顺序的逻辑关系。网络计划图一般包括单代号网络图和双代号网络图两种。Microsoft Project 2002默认的网络图为单代号网络图。 网络计划图有以下特点： （a）能全面而明确地表达各施工过程在各施工段上各项工作间的逻辑关系，使一个流水组中的所有施工过程及其各项工作组成了一个有机的整体（b）能对各项工作进行各种时间参数的计算，找出决定工程施工进度和总工期的关键工作和关键线路。 （c）能从许多可行施工方案中选出较优施工方案。 （d）能够迅速地根据变化后的具体情况及时进行调整，确保能自始至终地对计划进行有效的控制和监督，以达到降低成本的目的。 （e）网络计划的编制、计算、调整、优化和绘图等各项工作，都可以用电子计算机来协助完成，方便、快捷。 （4）单代号网络图（AON网络图） ?单代号网络图的表示方法 用一个节点表示一项工作，工作的名称、代号和持续时间均标注在节点内，用箭线表示各项工作之间的逻辑关系，这就是单代号表示法。 ?单代号网络图的组成要素 节点：一个节点表示工程施工中实际存在的一个工作过程(即一项工作)，其中也包括技术间歇过程。 箭线：每条箭线均表示各项工作之间先后顺序的逻辑关系。 线路：从起点节点到终点节点，沿箭线的指向，由一系列的节点和箭线构成的各条通路均称为线路。 [工具] 配置在Pentium Ⅲ，内存64M以上的电脑；装有Microsoft Windows操作系统及Microsoft Project 2002工作软件。 [地点] 机房 [教学内容] 1、在Project软件中输入项目的总体信息、如项目名称、起止日期等； 2、输入资源数据、建立资源数据库； 3、输入工作数据、建立工作数据库； 4、编制基本日历与资源日历； 5、调整项目相应的参数，观察对项目总工期和总成本的影响。 [学生任务] 项目计划 某工厂有一个机床大修的项目计划，其项目的详细信息如表所示：

牛津版高一英语必修一综合练习题

高一英语期中复习基础题选编（1） Ⅰ.语法填空 1．Bill’s mother was always telling him of the dos and don'ts，but it didn’t ________. A．use B．work C．stop D．promise 2．He was obviously annoyed，for he was not used to ________ like that. A．speaking B．spoken to C．being spoken D．being spoken to 3．Zhang Hua is clever and works hard at his lessons.________. A．So is Li Ming B．So does Li Ming C．It was the same with Li Ming D．So it is with Li Ming 4．I must have been a great________to my parents in those hard days，for I never failed to live up to their expectations. A．achievement B．comfort C．disappointment D．support 5．Good amounts of sleep every night ________also important for your health. A．is B．are C．was D．were 6．Even if I had taken part in the 400－meter relay race in the sports meet，our class________ the race.I’m ________ weight this year. A．would not have beaten；putting on B．might not have won；gaining C．cannot have defended our title；losing D．must not have achieved；acquiring 7．Teenagers ________ their health because they play computer games too much. A．have damaged B．are damaging C．damaged D．will damage 8．What you told me the other day is really true，________？ A．isn’t it B．isn’t what C．didn’t you D．wasn’t it 9．In the amusement parks，all the people love their jobs，________ might be not popular but are quite interesting regardless. A．which B．what C．that D．where 10．—Alice，you feed the bird today，________？ —But I fed it just now. A．do you B．will you C．don’t you D．didn’t you 11．In the past few years，My Heart Will Go On was a popular song among young people，________ were often heard singing it at parties. A．who B．which C．they D．that 12．________ is known to all，he is admitted to a key university. A．It B．What C．Which D．As 13．—Most of my classmates failed in the first exam.It seems there is no hope for us to make any progress.

人教版新课标高中英语选修7课文原文

人教版新课标高中英语选修7课文原文 Unit 1 Living well-Reading MARTY’S STORY Hi, my name is Marry Fielding and I guess you could say that I am "one in a million". In other words, there are not many people like me. You see, I have a muscle disease which makes me very weak, so I can't run or climb stairs as quickly as other people. In addition, sometimes I am very clumsy and drop things or bump into furniture. Unfortunately, the doctors don't know how to make me better, but I am very outgoing and have learned to adapt to my disability. My motto is: live One day at a time. Until I was ten years old I was the same as everyone else. I used to climb trees, swim and play football. In fact, I used to dream about playing professional football and possibly representing my country in the World Cup. Then I started to get weaker and weaker, until I could only enjoy football from a bench at the stadium. In the end I went into hospital for medical tests. I stayed there for nearly three months. I think I had at least a billion tests, including one in which they cut out a piece of muscle from my leg and looked at it under a microscope. Even after all that, no one could give my disease a name and it is difficult to know what the future holds. One problem is that I don't look any different from other people. So sometimes some children in my primary school would laugh, when I got out of breath after running a short way or had to stop and rest halfway up the stairs. Sometimes, too, I was too weak to go to school so my education suffered. Every time I returned after an absence, I felt stupid because I was behind the others. My life is a lot easier at high school because my fellow students have accepted me. The few who cannot see the real person inside my body do not make me annoyed, and I just ignore them. All in all I have a good life. I am happy to have found many things I can do, like writing and computer programming. My ambition is to work for a firm that develops computer software when I grow up. Last year invented a computer football game and a big company has decided to buy it from me. I have a very busy life with no time to sit around feeling sorry for myself. As well as going to the movies and football matches with my friends, I spend a lot of time with my pets. I have two rabbits, a parrot, a tank full of fish and a tortoise. To look after my pets properly takes a lot of time but I find it worthwhile. I also have to do a lot of work, especially if I have been away for a while. In many ways my disability has helped me grow stronger psychologically and become more independent. I have to work hard to live a normal life but it has been worth it. If I had a chance to say one thing to healthy children, it would be this: having a disability does not mean your life is not satisfying. So don't feel sorry for the disabled or make fun of them, and don't ignore them either. Just accept them for who they are, and give them encouragement to live as

最新牛津版英语高中必修一语法知识

第一讲定语从句 一关键词 定语从句：在复合句中修饰某一名词或代词的从句叫定语从句。一般紧跟在它所修饰的先行词之后。 先行词：定语从句所修饰的词叫先行词。 关系词：关系词用来引导定语从句。关系词有关系代词和关系副词之分。关系代词有that, which, who，whom，whose，as等；关系副词有when，where，why等。 关系词可以1) 代替先行词；2) 引导定语从句；3)在定语从句中担当一个成分。 例如：The car which my uncle had just bought was destroyed in the earthquake.句中，the car 为先行词；which是关系词引导是定语从句which my uncle had just bought，同时代替先行词the car在定语从句中作主语。 1.关系代词引导的定语从句 who指人，在定语从句中作主语。 whom指人，在定语从句中作宾语，常可省略。 which指物，在定语从句中做主语或宾语，做宾语时可省略。 that指人时，相当于who或whom；指物时，相当于which。在定语从句中作主语或宾语，作宾语时可省略。 whose 通常指人，有时也可指物，在定语从句中做定语。 Note: whose指物时，常用of which结构来代替： 关系代词在定语从句中作主语时，谓语动词要和先行词的性、数保持一致。 注意：which和as引导非限制性定语从句时，有时并没有特定的先行词，而是代表前面说的整个情况。这时which和as 被看作单数，译作“这一点”，其后的谓语动形式用单数形式。如：He said he was a Frenchman，which／as was not true. 但须注意，as引导的这种定语从句可置于主句之前，which引导的这种定语从句只置于主句之后。如：As is known to all，China is a socialist country. China is a socialist country, as/which is known to all. 2.关系副词引导的定语从句 when指时间，在定语从句中作时间状语。 where指地点，在定语从句中作地点状语。 why指原因，在定语从句中作原因状语。 注意：关系副词引导的定语从句经常可以用“介词+关系代词”引导的定语从句来表示。 介词+关系代词 1)根据动词选择介词。I don't know the foreigner with whom my teacher shook hands.（或I don't know the foreigner whom my teacher shook hands with.) Note: who和that不能直接位于介词后作宾语；有的短语动词是不能拆开的，常见的这类短语有：look for，look after等。

新课标人教版高中英语选修7单词(默写用)

选修7 Unit 1 1.disability n. __________ 2.disabled adj. __________ 3.hearing n. __________ 4.eyesight n. __________ 5.△syndrome n. __________ 6.△infantile paralysis __________ 7.△Rosalyn n. __________ https://www.sodocs.net/doc/5f4541654.html,p n. __________ 9.ambition n. __________ 10.ambitious adj. __________ 11.dictation n. __________ 12.△Sally n. __________ 13.noisy n. __________ 14.suitable adj. __________ 15.entry n. __________ 16.beneficial adj. __________ 17.△Marty __________ 18.in other words __________ 19.clumsy adj. __________ 20.△bump vi. __________ 21.outgoing adj. __________ 22.adapt vt. __________ 23.adapt to __________ 24.bench n. __________ 25.cut out __________ 26.microscope n. __________ 27.out of breath __________ 28.absence n. __________ 29.fellow adj. n. __________ 30.annoy vt. __________ 31.annoyed adj. __________ 32.△annoyance n. __________ 33.all in all __________ 34.firm n. adj. __________ 35.software n. __________ 36.sit around __________ 37.as well as __________ 38.parrot n. __________ 39.tank n. __________ 40.tortoise n. __________ 41.in many ways __________ 42.psychology n. __________ 43.psychologically adv __________ 44.make fun of __________ 45.encouragement n. __________ 46.conduct n. vt. __________ 47.△mainstream n. __________ 48.△fulfilling adj. __________ 49.never mind __________ 50.politics n. __________ 51.abolish vt. __________ 52.△abolition n. __________ 53.resign vi. & vt __________ 54.slavery n. __________ 55.literature n. __________ 56.△Barry Minto __________ 57.△Mount Kilimanjaro n. __________ https://www.sodocs.net/doc/5f4541654.html,panion n. __________ 59.assistance n. __________ 60.congratulate vt. __________ 61.congratulation n. __________ 62.bowling n. __________ 63.graduation n. __________ 64.certificate n. __________ 65.all the best __________ 66.architect n. __________ 67.△Sanders n. __________ 68.adequate adj. __________ 69.access n. __________ 70.accessible adj. __________ 71.△wheelchair n. __________ 72.handy adj. __________ 73.△earphone n. __________ 74.△impair vt. __________ 75.row n. vt . & vi. __________ 76.basement n. __________ 77.outwards adv. __________ 78.exit n. __________ 79.meet with__________ 80.approval n. __________ 81.dignity n. __________ 82.profit n. __________ 83.△italics n.（pl） __________

项目管理与Project的应用

项目管理与Project2000的应用 一、项目管理与Project2000的关系 在项目管理知识体系中，包括九大知识领域（范围管理、时间管理、费用管理、质量管理、人力资源管理、沟通管理、风险管理、采购管理、综合管理）。但从项目管理辅助工具软件Project2000来说，它包含了这九大知识领域中的5大核心领域，另外4个领域需要通过其它辅助工具或人工操作来完成。包括的5大领域如下所述： 1、范围管理 项目管理的第一个知识领域就是“范围管理”，在项目范围中，包括两层内容：一是项 目范围；二是产品范围。项目范围针对的是我们项目的目标，包括软件开发、集成、培训和项目实施等。产品范围侧重在软件的需求范围，可以理解为对项目范围的一个重要补充。两者有一定关联，但也各有侧重，只有这两者相加，才构成了我们完整的项目范围。然而在Project2000工具中，指的项目范围是指第一种,结果输出就是WBS分解。 2、时间管理 时间管理也称进度管理，在Project2000中，它提供了工期估计、工作搭接关系、进度安排、进度控制等基本功能。还能够自动计算出关健路径，可以很方便的设置我们的里程碑控制点，并能够实现项目的动态跟踪。还提供了多种时间的管理方法，甘特图、网络图、日历图等。应该说时间管理，是Project 2000中最强大的功能。 3、费用管理 在费用管理中，Project2000中采用的是“自底向上费用估算”的技术，由于它是依赖每个WBS任务的估算，所以使得费用估算更为准确。并且它还能与EXCEL等进行结合，生成费用曲线图和挣得值趋势图。

4、人力资源管理 在人力资源管理中，Project2000提供了人力资源的规划、人力资源责任矩阵、资源需求直方图、资源均衡等，它能帮我们做好资源的分配、进行资源的工作量、成本和工时的统计。 5、整合管理 项目管理的整合管理就是对于整个项目的范围、时间、费用、资源等进行综合管理和协调，在Project2000中，它能根据范围、时间、资源的变化自动进行相应计算和调整。 二、 Project2000使用前的环境设置 在进行计划编制前，需先设置好Project2000的使用环境，这样可以更方便我们计划编 制时的操作。环境设置一要根据自己的习惯，二是要根据项目的实际情况。下面介绍几种常用的环境设置项。 * 首先，要设置项目摘要信息，在摘要信息中需输入该项目的标题、项目经理和单位等信息。主要是便于打印时显示的信息。 操作方法：选中菜单文件－》属性－》摘要信息－》。 * 设置项目的日历，默认从星期日开始，中国人的习惯一般是从周一开始。 操作方法：选中菜单工具－》选项－》日历－》每周开始于－》。 * 设置任务类型，默认为“固定工时”，固定工时的含义是当一项任务分配给一个人做是10天，当增加一人时，则工期自动变为5天。我们的操作习惯应该是人员增加时，原工期一般要求仍不变。所以需选择为“固定工期”。 操作方法：选中菜单工具－》选项－》日程－》默认任务类型－》。 * 设置WBS编号，默认无WBS编号，为了计划阅读清晰，建议设置大纲编号作为WBS编号。 操作方法：选中菜单工具－》选项－》视图－》选中显示大纲编号。

2019-2020年牛津译林版高中英语必修一Unit 1《School Life》(Project)教案

2019-2020年牛津译林版高中英语必修一Unit 1《School Life》 （Project）教案 【教学目标】 1.to arouse students’enthusiasm for developing after-class activities and forming a school club; 2.to gain the general idea of how to make an attractive poster for the club; 3.to strengthen students ability of putting theory into practice; 4.to guide students to cooperate effectively through group work. 【教学重点】Have students discuss and learn to finish a project by working tog hter 【教学难点】Students should search and find information, and do some writing and drawing by themselves 【教具】Multi-media projector Are you impressed by the soft background music? Step 2 Attract Your Eyes & Grasp Your heart Do you want to experience

(2) Tom in sch ool. e than five courses this term. Jack is no more diligent than John. 杰克和约翰都不勤奋。

最新人教版高中英语新课标必修一单词表

高一英语人教新课标必修1重点单词词组归纳总结 必修1 Unit 1 重点单词 1. add vt.增加；添加；补充说vi加；加起来；增添 2. upset vt&vi.使不安；使心烦adj.心烦意乱的；不适的；不舒服的 3. ignore vt.不理睬；忽视 4. calm adj.平静的；镇静的；沉着的vt.&vi.（使）平静；（使）镇静 5. concern vt.关系到；涉及n. 关心；关注；（利害）关系 6. cheat n.欺骗；骗子vt.&vi.欺骗；骗取；欺诈；作弊 7. list vt.列出 8. share vt.分享；均分；分担n.一份；份额 9. series n.连续；系列 10. crazy adj.疯狂的；狂热的 11. purpose n.目的；意图 12. dare vt.&v.aux. 敢；胆敢 13. thunder n.雷；雷声vi打雷；雷鸣 14. entirely adv.完全地；全然地；整个地 15. power n.能力；力量；权力 16. according adv.依照 17. trust vt.&vi.信任；信赖 18. suffer vt.&vi遭受；忍受；经历. 19. questionnaire n.调查表；问卷 20. quiz n.测验；提问vt. 对…进行测验 21. situation n.情形；境遇；（建筑物等的）位置

22. communicate vt.交际；沟通；传达（感情、信息等） 23. habit n.习惯；习性 重点短语 1. Calm down 平静下来；镇定下来 2. Be concerned about 关心；挂念 3. Make a list of 列出… 4. Be crazy about 对…着迷 5. According to 根据…所说；按照 6. Get along with 与…相处；进展 7. Fall in love 相爱；爱上 8. Try out 试验；试用 9. add up 合计 10. set down 放下；记下；登记 11. get sth. done 做…；使…被做； 12. share sth. with sb. 和某人分享某物 13. go through 经历；经受； 14. a series of 一连串的；一系列；一套 15. on purpose 故意 16. in order to 为了… 17. join in 参加；加入 18. communicate with 和…交流 19. face to face 面对面地 20. suffer from 遭受。。。 必修1 Unit 2 重点单词

牛津译林版高中英语必修一一、 用单词的正确形式填空

一、用单词的正确形式填空 1. The trip was much more ________(enjoy) than I had expected. 2. We prefer some more ________(experience) workers to work for us. 3. He tried many ways of ________(earn) money and in the end he became a businessman. 4. The new dress makes her more ________(respect). 5. ________(devote) herself to her family, she felt she had lost herself. 6. Much to his ________(satisfy), his son was admitted to Peking University. 7. My husband does all the ________(cook) at home, which makes me very pleased. 8. Thanks for your advice and ________(encourage)． 9. He found his new job a little ________(challenge)． 10. The students are working hard every day, in ________(prepare) for the big examination. 二、选择适当的单词或短语填空 1. use to/be used to (1) He ________ going to bed at 10 o'clock every night, which was good for his health. (2) Today's children are not like what we ________ be. They are much more confident. 2. because/because of (1) The sports meet was finally put off ________ the bad weather. (2) Many people do exercise every week, ________ they have realized the importance of good health. 3. prepare for/prepare (1) Last Sunday, I gave our house a thorough cleaning and then ________ a wonderful meal and cooked a special dish for my wife. (2) When people moved to a new country, they have to first ________ the new surroundings(环境)． 三、根据汉语提示完成句子 1. 我丢了那本红封面的书。 I lost the book ________ ________ is red. 2. 对东京来说，赢得2020年夏季奥运会举办权是一件非常成功的事情。 ________ ________ ________ ________ ________ for Tokyo ________ ________ the right to host the 2020 Summer Olympic Games. 3. 一完成家庭作业，他就去打篮球。 ________ ________ his homework, he went to play basketball. 4. 她是一位如此受人尊敬的老师以至于我们都爱她、尊重她。 She is ________ ________ a teacher ________ all of us love and respect her. 5. 经历这不同的生活方式我很幸运。 I was very ________ ________ ________ this different way of life. 四、单项填空 1. She devoted all her time and energy to ________ the little child. A. look after B. looking after C. looked after D. looks after 2. While you are away, we will keep you ________ of the latest development of the project in time.

使用Project进行项目管理

实验六使用Project软件进行项目管理 （一）实验目的 1）理解计算机在IT项目管理中的作用； 2）熟悉Microsoft Project 2003的界面； 3）掌握Microsoft Project 2003的常用功能和操作。 4）学会使用Microsoft Project进行时间管理、资源规划、成本管理。 5）学会使用Microsoft Project跟踪项目进度、跟踪项目资源状况、跟踪实际成本。（二）实验背景知识 项目管理软件是辅助项目管理人员完成项目动态管理、项目协调调度以及系统集成控制、完成工程的质量、进度、成本控制，完成项目合同、人力资源的管理，以及完成项目费用估算、风险分析、不可预见费用估计等。 项目管理软件的作用： 1、根据项目生命周期确定管理流程（见图1和图2） 图1 项目生命周期各阶段的管理流程

图2 使用Project进行项目管理的流程 2、Project自带的项目模版 （三）实验内容和步骤 一、初步实验 以下为一个新产品上市的营销项目，根据以下信息利用project 2003实现项目的创建。

任务名称工期开始时间前置任务准备阶段11 工作日2010年7月1日 企业评析7 工作日2010年7月1日 市场分析7 工作日2010年7月1日 竞争分析7 工作日2010年7月1日 举行筹备会议 3 工作日2010年7月12 日2，3，4 讨论会议内容 1 工作日2010年7月15 日 5 企划阶段17 工作日2010年7月16 日 1 问题与机会7 工作日2010年7月16日 销售预算与目标7 工作日2010年7月16日 产品目标7 工作日2010年7月16日 定位策略 3 工作日2010年7月27 日8,9,10 目标市场 2 工作日2010年7月27 日8,9,10