CN101930441A - Method for establishing Internet holographic search service - Google Patents

Abstract

The invention relates to a method for establishing Internet holographic search service and belongs to the field of distributed type, in particular to an application of a variety of large-scale distributed technologies on the improvement of Internet. The method comprises the steps of utilizing intellectual resources serving an Internet application system; using the intellectual resources to control the Internet application system; establishing a distributed calculation environment, that is a large-scale non-central Internet under the control of the intellectual resources; reusing the current Internet resources under the control of non-intellectual resources; ensuring sufficient computational resources for the search service through the participation of all the computational resources; searching under the guidance of the intellectual resources; and realizing the main characteristic of holographic search. The method can realize a holographic search system with abundant computational resources for support, real-time property, high intelligence, high performance-cost ratio, large data size, large particle size, stateful property, intellectual resource leading, large coverage rate, time characteristic, strong personality, multi-format property, strong people-oriented property and strong interaction.

Description

A kind of method of setting up the internet holographic search service

Technical field

The invention belongs to the application of distributed computing technology on the internet direction.The details that wherein relates to comprises that distributed route technology, distributed type assemblies technology, the Internet-based should be with technology and Internet social network systems etc.

Background technology

In the middle of distributed system [1], search technique is a kind of route technology [2] in essence.So-called route technology has special definition in the middle of distributed system, promptly in the computing environment that a plurality of computing equipments constitute, under the different prerequisite of physical location, seek satisfactory computational resource, thereby satisfy the demand of certain computational activity to resource.In the large-scale non-central Internet internet, more wide in range to the definition of computational resource, promptly computational resource is not limited to the resource that computing equipment itself is had, and as CPU, memory space, input and output and bandwidth etc., also comprises the intellectual resources of controlling these resources; Perhaps both are separated, be called computational resource and intellectual resources.Such division intention is for the large-scale non-central Internet internet, because wherein main computational resource is all under intellectual resources control independently, effective utilization of considering computational resource merely is really to reach the high performance purpose of system; Simultaneously, even in the so-called large scale system, resource also is rare forever.In the process of configuration scarce resource, must take the situation that intellectual resources show computational resource into consideration, otherwise any resource distribution mode of runing counter to therewith can not work.

Search technique is the comprehensive results that use of multiple technologies.What high-quality search technique at first will be considered is the concrete computing environment that faces.The computing environment here can be divided into local search, center type search [2] and non-central formula search [2] by distributed model.This classification has embodied the difference on the physical location between the computational resource that participates in search and the searched resource.Computing environment can be divided into text, picture, audio frequency and video by the object search classification.Because the restriction of current computing technique, though object search has multiple difference, actual search is based on also in essence that text carries out.In addition, from the hunting zone, search technique can be divided into horizon scan and vertical search again.So-called horizon scan refers to the search of full field, and promptly the hunting zone is not limited to a specific field.Vertical search in contrast is promptly at the search of specific field; If this be since search technique can with the knowledge feature combination of specific field, can improve search efficiency greatly.Classify according to the search means, mainly comprise keyword search and natural language understanding search [3].The restriction of current computing technique remains main flow based on the search of key word.Also having a kind of classification is that the mode of machine perception is searched for; Owing to relate to some insoluble problems such as computer intelligence, the conceptual phase that these modes also stop.

The main feature of current internet search technique at first is the search service that is subject to limited resources.Since the drawback on the current internet basic structure causes whole internet, applications to be moved under the relatively limited condition of computational resource, this search technique that certainly will cause setting up thereon also is to work in the face of under the situation of resource shortage.Computational resource is that computing system provides the prerequisite of function, performance and even intelligence to guarantee; Under the assurance that does not have enough resources, search technique also seriously is limited by resource bottleneck.In order to address this problem, provide the supplier of search service effectively to utilize the limited resources except using multiple complex distributions formula technology, must take to drop into expensive means and remedy deficiency on the resource.But even so, facing magnanimity internet data and numerous search subscriber, resource remains serious bottleneck.

Current internet search technique right and wrong are searched for immediately.Since internet hunt is positioned to search service for the user provides whole internet data, have to consider the feature that searched data constantly change so.This variation is right and proper.The way that current internet search service supplier takes is that at first each website from the internet obtains all possible datas, and on the server of own foundation the data of obtaining is carried out index and ordering.After receiving the request of user, in the index database of having set up, mate according to this key word based on key word; Simultaneously matching result is returned to the user according to prior clooating sequence.In the middle of such search mechanisms, be recent renewal in order to ensure data, the search supplier must hold period obtains the possibility of data, promptly constantly obtains latest data from the internet.Yet because internet scale is huge, this process of obtaining data can't guarantee instantaneity.For fear of such mechanism the user is caused damage, the supplier of search service can only accomplish on the distributed arrangement of search server that scale is big as far as possible, thereby has more strong big computational resource to obtain latest data with the minimum delay.Data are effective but this effort is for upgrading relatively slowly; For upgrading new data frequent and that continue to bring out, keep the difficulty of instantaneity big.

The current internet search technique is intelligent limited search.Search technique requires high to so-called intelligence computation [4].Such requirement is mainly reflected in understanding to searching request, to understanding and the matched rule in arrangement, searching request and the searched data and the aspects such as ordering of Search Results of the analysis of search subscriber historical record, searched data.The solution of these problems relates to the bottleneck of computer science itself, i.e. the machine intelligence problem.It may be irrealizable attempting to produce intelligence by machine algorithm.In the face of such difficulty, some search techniques attempt utilizing intelligent resource to solve the indeterminable problem of machine [4].So-called intelligent resource refers to user behavior and the state that exists in computing system.These information come from the user and come from the people who handles or use computing system in other words; Anyone works under intelligence domination, and the reaction that this behavior produces during at different data or system state is different.By investigation, can be familiar with to some extent corresponding data or system state to human these behaviors and reaction.This understanding is applied to can produce useful influence to the algorithms that need intelligence to help in a large number in the middle of searching in the middle of the search.These principles have obtained utilization in current search, as PageRank[5] etc.But because the restriction of current system limited resources has caused obstacle to collection, control, these intellectual resources of management and use.Server is tired in response user's request, does not have the processing of residual resource assurance to above-mentioned intellectual resources.This has fundamentally limited the intelligent characteristic of search.

The current internet search technique is low cost performance search.Strictly speaking, current internet search is low cost performance search, and promptly current the prerequisite of search service is provided is except appropriate solution, also the mode that must possess abundant computational resource and use these computational resources.Obtaining the data procedures from the internet, abundant computational resource is the assurance that promptly obtains the magnanimity internet data relatively; When the response searching request, abundant computational resource is the prerequisite that makes that the user avoids waiting as long for.But even so, be based upon that high-performance in the high investment also just carries out at the lightweight data; In case data characteristics changes, the problem of its resource scarcity can highlight again.

The current internet search technique is the lightweight search.On the present Search Results of lightweight mark sheet of current internet search technique.The supplier of internet data and service search at first is the gleaner of an internet data and service.Though internet data and service are huge, specifically the data of each hyperlink correspondence are small from data volume.To be that computational resource is limited cause its essential reason.The Search Results that produces based on same structural the Internet search server can not become big data quantity, also can not combine them and forms a bigger data set even search server obtains numerous Search Results.In fact, the current search server does not even provide real data, and just points to the address of data, promptly the burden of obtaining data is handed to the original server that this address is pointed to.The result who does like this is that the search server burden reduces greatly.Some search server can provide the buffer memory service, thereby avoids original server to visit and make the user can not get Search Results.And such buffer memory also is to be that unit carries out to the data of network address correspondence only, can not carry out any combination that causes data volume to increase to the data that search.

The current internet search technique is the small grain size search.Data on the current internet at first are that lightweight is a feature.When combining between a plurality of lightweight data, will produce the notion of granularity.Granularity refers to the linear module of data implication coverage rate.The small grain size feature and the lightweight feature of current internet are closely related, and just go to be familiar with same problem from different angles.Current internet data is combined into the certain layer aggregated(particle) structure according to relation intrinsic between data usually.When the such data of user capture, in order to reduce the pressure to server, the user can only obtain the data of a node on this hierarchical structure, i.e. small grain size data.Based on such internet system, any improvement is not done in search service to this.As search server, the intrinsic relation between the small grain size data that search is not carried out any combination, thereby form the bigger data of granularity, but keep the state of its small grain size.Such way also is unfavorable for the expressed intact of Search Results.In addition, small grain size is searched for even can be had serious consequences to the current search service.On the one hand, internet data must be constantly duplicated in search service, and this is the work that must finish as the internet hunt ISP; On the other hand, because data scale constantly increases, cause the user when height mates the small grain size searching request, could obtain satisfactory result again.This means that the current internet search service has entered a vicious cycle: necessary constantly copy data, and the increase of data volume can cause the Search Results overexpansion, numerous valuable small grain size data are submerged wherein; Final result is that search service has been degenerated to simple match service; The PageRank technology effect in the face of huge data source time that is used for weighing the quality of data is more and more littler.

The current internet search technique is the stateless search.Stateless also is the feature of current internet, and this is the strategy of taking in order to guarantee limited resources to offer more potential users equally.State in the search maintains and is beneficial to the intellectual resources of utilizing search subscriber to have.The user when search possible state comprise the key word of current search, repeatedly have or not between the search contact and and other search subscriber between have or not mutual etc.The supplier of search service can utilize these status datas in the follow-up service process, improve the search service quality.The current search technology is not done any maintenance to searchers's state, does not perhaps utilize these status datas to contribute for search in search.

The current internet search technique is the search that machine algorithm is taken as the leading factor.Search is to need most the application that embodies intelligent characteristic on the internet.But there are two deficiencies in the current internet search technique on intelligent problem, promptly or pursue machine algorithm and produce intelligence [4] or do not make full use of intellectual resources [6] in the system.The former problem is actual to be the notional mistaken ideas of computer science, and the latter then can not be shared by most of users and causes owing to computational resource.The current internet system is a kind of distributed computing platform of high concentration.One of feature of this model is exactly the leading main computational resource of machine, and user people in other words is in aided location.The strict limited subscriber of machine algorithm takies computational resource, as far as possible by scheduling, reduce multiple means such as data granularity, stateless and control limited computational resource, thereby dealt with potential user capture.This computation model of taking as the leading factor with machine algorithm derives from the conservative algorithm that the deficient state of computational resource produces down.Though this algorithm has protected computational resource too much not taken by a user to greatest extent, its cost also is that significantly promptly the intellectual resources of user behavior representative can only contribute to system limitedly.The current internet algorithm embodies the re-computation resource in this sense and ignores the thought of intellectual resources.When intellectual resources showed as the user to the taking of computational resource merely, this algorithm was effective; Current internet calculates that such theory sets up with regard to being based on.And when intellectual resources except taking computational resource, when also playing the part of computational resource supplier's role, be that the algorithm of feature just has been false to ignore intellectual resources in the past.The algorithm of this new thinking applies to search for when central, and the algorithm that intellectual resources are ignored in the search that machine algorithm is taken as the leading factor in other words just must obtain changing.

The current internet search technique is low coverage rate search.For internet hunt service, the search coverage rate refers to data and the service that search server contains and accounts for the ratio of whole internet datas and service.The calculating of this ratio can be compared with the storage capability on the Global Internet server by the storage capability that search server had and draw a rough result.Why be rough result, because be all to have the massive duplication data on the search server or on the Internet Server, and internet data is all increasing constantly.But can think that this ratio is little.Certainly, general search server can not preserved complete internet data, and is index.Even but like this, this ratio can be too not high yet.General estimation Google[7] can reach 10%[8].Google is attempting to duplicate the data on the whole Internet Server, in the hope of more perfect search service is provided.This way can think to machine algorithm extreme supertition caused, in other words conj.or perhaps to the extreme challenge of human mind's resource.From pure physics storage angle, perhaps this target can realize; Let us not go into the question now, and whether this overlapping investment is necessary, if Google is in commercial success like this, not as allowing it provide Internet Server for the whole world.But data and service are not simple literal, picture, Voice ﹠ Videos, and it also comprises generation, obtain, use, change, innovate, duplicate even the people's of deleted data and service state, thought, inspiration and mood, and these things can't duplicate.When these factors can become computational resource number leading, temporary transient resource inferior position can since the intervention of these factors make to blindly supertition and the way of intellectual resources ignorance is met one's Waterloo of computational resource.

The current internet search technique is no time search.Data on the internet and service much possess temporal characteristics, i.e. data and service is along with the variation meeting of time constantly changes.Secondly, body is because of the time changes even the data kimonos is attended to the basic or the fundamental, and along with the appearance of similar new data and service, old data and service meeting are progressively out-of-date and be eliminated, but can have some data simultaneously and service is long-term keeps high-quality and do not lose value.Once more, some data also has the period value with serving, and promptly has special value in some periods, and this period is worth or decline or maintenance even lifting and surpass.At last, the internet is a system that constantly changes, and the generation of data and service has instantaneity.Can a search system obtain so instant data and service and be presented to the user also is the problem that search technique need be considered.The current search system data and the service timeliness solution on towards long-term valuable data.Owing to lack the instant ability that obtains data, in fact the current search service is not consider time factor when data sorting or management.And,, promptly website main but limited quantity is monitored separately, in the hope of obtaining up-to-date more new data for generally having taked the special treatment method of artificial interference with Time Correlation Data in view of the importance of time factor; And when these data and user search request appearance coupling, search system can be given right of priority in ordering.This method is actually it is mainly sorted and the simple modifications of data managing method, to remedy the deficiency of system on timeliness.But generally speaking, the current search algorithm is not based on the search of time.

The current internet search technique is non-personalized search.Personalized search also is a reluctant problem in the middle of the search system.Owing to be difficult to obtain user's real demand, can't make different processing to user's request.This is equally because the computational resource deficiency causes, even promptly can't follow the tracks of or carried out following the tracks of and can not make a response according to this tracking results in the limited time user behavior.In addition, because the stateless feature of current search system, also feasible understanding to user's request becomes not enough.At last, personalization is a complex conception.It is the unique need of user's individuality, also is associated user group's common trait; If any individual character does not have general character, do not exist in all senses yet, the data and the service that promptly do not have to satisfy its demand exist.Based on the analysis of individual consumer's state, calculated amount is big, also enough not reliable algorithm.With it limited computational resource is used for each user's judgement, not as satisfying individual character by the assurance of general character by suitable method.This had both reduced the pressure that calculates, and also can determine a sexual demand more accurately.In fact, the current internet search technique can also be called does not have classified search, and the data that promptly copy to self system from the internet can't effectively be classified to it.This also is the another reason that causes current internet with no personalization.The current search system is moving towards to duplicate the awkward predicament of whole internet, after obtaining mass data, owing to can not effectively classify to data, all users can only extremely simplify the data that bulky complex is observed at visual angle (as famous Google simplifying interface) from a unification.It is to obtain to mass data that the search visual angle of superelevation granularity and the data access patterns of small grain size in fact cause the user.If coverage rate is with regard to low search service originally, because above-mentioned reason makes the user can only see finite data, the difference of its search quality just can have been infered again.In fact, when using the current search system, have only the enough little key word of searching request granularity just can obtain relative satisfactory result; And the searching request granularity is when excessive, and the Search Results quality can seriously descend.The personalization that this means current search is decided by whether searchers's searching request is personalized, rather than whether search technique provides personalized function.

The current internet search technique is the single search of content.So-called content is single to refer to the undue homogeneity of Search Results, promptly in the data layout based on text, and relevant picture, Voice ﹠ Video data will be lacked a lot.This carries out owing to current search also is based on text on the one hand, and is immature at the searching algorithm of extended formatting data; On the other hand, current internet lightweight data are main feature, also cause except that text, and the extended formatting data are limited.At last, after the data of different-format were carried out polymerization, system also was difficult to bear the pressure that the heavyweight data cause server.

The current internet search technique is data-oriented search.With the corresponding way of search of data-oriented search be search towards people or intellectual resources.Any data and service are all created by the people and are provided.When obtaining search data and service,, can more meet user's request so if can obtain the data of relevant people simultaneously.People's search is actually the origin of data and service is done search.Any one data or service providers all do not have special computing equipment in order to data and service to be provided in the current internet, a people can issue on a plurality of servers, and a server can bear data and the service that a plurality of people issue, with regard to for the management the whole enterprising pedestrian in internet causes difficulty, be difficult in and set up simple corresponding relation between people and the data like this.This causes obstacle from data-oriented to the transition towards the people for search.

The current internet search technique is no interactive searching.The consequence that data-oriented search technique is brought is exactly the no interaction feature in the search procedure.Here refer to mutual between the user alternately, rather than mutual between user and the search server.Also exist at searching algorithm itself under the situation of defective, if search system can be assisted people's search, and can set up contact between searchers and the data author, promptly directly link up mutually, so such search pattern can be more effective.

This section has been done summary to the current internet search technique, has focused on the weak point of current internet search technique.And the reason of these weak points to conclude be exactly that the shortage of computational resource causes intellectual resources are utilized limited.Be based on the recognition, when improving above-mentioned deficiency, should concentrate the limited problem of computational resource that solves.The solution of this problem can not solve by the simple approach that increases computational resource merely again, and the such mode of one side drops into excessive, and this on the other hand simple mode can not form the fusion with intellectual resources.Rational solution is to be that intellectual resources enter system swimmingly and provide convenience when increasing computational resource, thereby search service can be carried out on this basis of computational resource that intellectual resources are controlled.The method that only has above-mentioned two characteristics concurrently can fundamentally overcome the problem that the current search technology exists.Fig. 1 is the simple expression of current internet search mechanisms.

Summary of the invention

Creation purpose of the present invention is by improving internet, applications layer basic structure, increase the internet computational resource thereby reach, make intellectual resources can be effectively control these computational resources simultaneously, finally on such basis, set up brand-new internet hunt service according to self wish.Fig. 2 has represented the operating mechanism of holographic search.

The present invention realizes goal of the invention by following technological means:

The first step is set up low-cost search service mechanism.Search service is provided, especially so-called parallel search is very high to the requirement of computational resource.It is big that its reason comes from its potential user capture amount, and need the data volume of covering big.Such requirement all needs high investment concerning any one search supplier.Search solution provided by the invention has changed the source of resource, makes system resource obtain more computational resources under the situation that need not extra input and supports search, and this is the main cause that forms low-cost search service.The another one reason that forms low-cost search is the effective utilization to intellectual resources.Machine algorithm need be paid a large amount of computational resources when carrying out so-called artificial intelligence, and poor effect.Search service is to the requirement height of intelligence computation, and if have enough intellectual resources to simulate the computational resource of intellectual resources so will be saved in the system for utilizing, thereby further reduce the cost of search service.

In second step, set up instant search mechanisms.So-called instant search refers in the search procedure ability to the high data immediate reaction of change frequency.Here comprise two parts problem, comprise data and Service Source and existing internet data and Service Source in the large-scale non-central Internet internet respectively.Because both are in the middle of the different environment, processing mode is different.For the former, because interviewee's each node in the large-scale non-central Internet internet all is to work in the mode of initiatively expansion, when searched person's related data changed or have new data or service to occur, change or new data can be sent in the mode of initiatively diffusion in the cluster on all nodes.Instant search is finished in the large-scale non-central Internet internet easily.Will accomplish this point for mass data on traditional internet and service, what at first will consider is that the conventional the Internet service device is transformed, and promptly reaches the conversion of conventional the Internet service device in behavior by paying computational resource.This computational resource of paying is mainly used in periodic access conventional the Internet service device to obtain its data and service new or that change.Simultaneously, for reducing the pressure of this part resource, can also reduce the demand of periodic access by the access ability of utilizing node in the large-scale non-central Internet internet to resource.By the way, equally also can reach the purpose of instant search.

In the 3rd step, set up strong intelligent characteristic search mechanisms.In the middle of the large-scale non-central Internet internet, intelligence is embodied on the accumulation of system's intellectual resources and utilizing, rather than utilizes machine algorithm to obtain to be similar to human intelligence.The core difference of large-scale non-central Internet internet and traditional internet system just is that in fact the former function, performance and intelligence depend mainly on the central user's of system behavior, and the latter's function, performance and intelligence depend mainly on machine algorithm and computational resource.Because user behavior has determined combination, quality, life cycle of computational resource or the like, will make judgement to this status in user's intelligence brains of computational resource conversely to the monitoring of these states.The ability that computational resource shows when carrying out such monitoring is that the people is not available.Finally, consequent intelligence can act on the user again, makes user's generation system possess the understanding of intelligent characteristic.This interaction must be to carry out under the prerequisite of equal opportunity, is in promptly that Any user all has identical chance to show its intellectual resources in the system.Based on above-mentioned multiple understanding, holographic search technique provides the search of strong intelligent characteristic, mainly show ranking results rationality, related data and service provide and the search of correlation computations resource on.

In the 4th step, set up the high-performance search mechanisms.The performance of search also is one of search service important indicator.Search technique of the present invention is guaranteeing that low relatively computational resource drops under the prerequisite, still can keep the high-performance of searching for.When general search subscriber was searched on the internet, can stand the stand-by period all can be very short.If can not guarantee the performance of search service, other features can not accepted by the user yet so.In addition, high-performance except be embodied in rapidly user search request is made a response, also be embodied on the processing power of counterweight magnitude data.The not responsible transmission to data usually of tradition internet hunt mechanism, the user need pass through the direct visit data of hyperlink.Current, some search services are carried out buffer memory to internet data, and provide data transport service for the user.But when facing the heavyweight data, performance is still not enough.The transmission of heavyweight data can be effectively handled in search of the present invention.

In the 5th step, set up data-oriented heavyweight search mechanisms.The current internet search service is not data-oriented search, perhaps is called the search service towards hyperlink.User's searching request obtains is not real data but hyperlink, promptly indirect result.The Search Results of this form influences for the lightweight data not quite; Hyperlink result is similar to the index to data.For the heavyweight data, the user has to go to obtain related data by hyperlink, and the limited in one's ability of heavyweight data handled in this process because traditional internet, and period of reservation of number is long.In the middle of holographic search technique, owing to be data-oriented, user's Search Results directly shows the user with the form of final data; Under the support of high performance system, the heavyweight data also can continue to be transferred to the user.

In the 6th step, set up the coarsegrain search mechanisms.Granularity has reflected a logic node in the middle of the structure of knowledge.Coarsegrain is represented the more complete structure of knowledge.The service of tradition internet hunt shows deficiency on the search granularity, promptly the granularity of Search Results is little.Search of the present invention provides the coarsegrain search to satisfy the demand of user to complete Search Results.No matter be in the face of data and service in the large-scale non-central Internet internet, still in the face of the data and the service of traditional internet, holographic search has all solved the problem of current internet search small grain size.

In the 7th step, establish state search mechanism.The state of recording user has huge help to the intelligent characteristic that improves search service in search procedure.The current internet search is to have inherited traditional internet, applications stateless feature, promptly search for the supplier and be indifferent to residing on the internet state of active user and intersection record in the past, only the key word that provides according to the user is searched in unified data area.Such result is exactly same key word to all users' result all is the same.This does not meet the diversified demand of search subscriber.The user generally comprises following state in search: visit in the past and searching record, current search request, current channel of living in, channel correlation, current and other user interactions situations and in the past with other user interactions records or the like.Search service can be carried out the intelligent search service to the current search request according to these states.According in the past the visit and searching record in the associated user, can provide potential related data sources for the user; Should continue to preserve for the current search request, like this when related data and user change, can instant and search subscriber communication; According to the current channel of living in of user, relevant data in the channel and service can be offered the user; In the holography search, the data that search out are relevant with channel, can obtain the data and the channel itself of a plurality of channels usually, and these states have been done assurance for system continues to provide related data; Because the residing computing system of the present invention is strong alternately, and thisly also can user's search be worked alternately, so interaction mode also is used to auxiliary the search.The effect of these states in search also with legacy system in different, promptly be not to be so-called machine intelligent system service by these states, but by these states find more heterogeneous pass and available intellectual resources serve the search and the searchers.In order to achieve the above object, system need provide powerful condition managing means and corresponding calculated resource.

In the 8th step, set up the high coverage rate search mechanisms.For the data on the multiple internet more can searched service system be covered, thereby, improve the search coverage rate and become one of search system key issue for the user provides more complete Search Results.In the large-scale non-central Internet internet, the coverage rate of search can obtain much higher data coverage rate by the cost that is lower than traditional approach.

In the 9th step, set up the search that possesses temporal characteristics.Any data all can be related with the time.This contact can be expressed as long-term time validity feature, break time validity feature and instant time validity feature roughly.So-called validity feature refers to and is satisfying under the prerequisite of time conditions, and data just can possess meaning or could present to the user as Search Results.Search mechanisms of the present invention can for the user provides the Search Results of different time attribute, thereby make Search Results more complete on temporal characteristics according to user's unified searching request.

In the tenth step, set up personalized search mechanism.Personalized search also is fully reflected in the middle of search mechanisms of the present invention.In the large-scale non-central Internet internet, personalization can more be broadly defined as overall situation personalization, group is personalized and individual personalization.So-called overall personalization refers to the search need that all searchers may possess a certain searching request.In fact the current search service belongs to such type.Search technique of the present invention also can be finished such task.The personalization of so-called group is the real demand to some searching request that comes out in certain local reaction to only.This has also reflected a kind of universal demand in the middle of the search.But the current internet search service causes also not being difficult to the search crowd is done any identification, division and processing at computational resource and to the deficiency on the intellectual resources understanding, can't accomplish group's personalization.As holographic search technique, under the reliable prerequisite that guarantees of computational resource,, reach satisfying of group's personalized search by the leading group of intellectual resources.Individual personalization is a kind of more extreme personalization.Strictly speaking, any individual personalization all is and a certain group even personalized relevant with the overall situation; Only under extremely special situation, just can there be an absolute personalization.But in the system that intellectual resources are taken as the leading factor, the individual character that can occur a few peoples usually is leading most of people; And most individual characteies that are affected difference slightly.Holographic search system attempts to reach the purpose of individual personalized search by enriching under the computational resource support.This target is powerless especially to the current search service.

In the 11 step, set up the multi-format data search mechanisms.Expression to same things or meaning has multiple mode, from literal, picture to audio frequency and video.For this situation, search service should feed back to the user to the mode with Search Results after these data combinations.Search mechanisms of the present invention is mutual according to the standard of user's distributing data form and user's, has realized the search service of multi-format.

In the 12 step, set up search mechanisms towards the people.One of principal character of holographic search is exactly that Search Results not only comprises data, also comprises creator and other relevant crowds of data and service.Holographic search thinks that the search to the people is only the final purpose of search: after obtaining to produce the source (being the people) of data and service, just have further personalized interaction until new data and creator's appearance newly.

In the 13 step, set up interactive search mechanism.In search, obtain under people's the prerequisite, by the interaction technique in the large-scale non-central Internet internet, can assisted user in search procedure with the supplier of raw data, carry out similarity other searchers and with the relevant reader of this search data and other creators carry out based on various member's scales, various data scales alternately.This should be the user obtains valuable data in search strong ancillary method alternately.

In the 14 step, set up holographic search mechanisms.Why search technique involved in the present invention is called " holographic search ", and its intention at first is meant the integrated degree of search data, provides other possible related datas except matched data as this search service, also wants high on the data level of coverage; Secondly, holographic search represents that also search data itself is also abundanter in nature, as Search Results being comprised multiple form, possessing timeliness and instantaneity, can also hold variation characteristic or the like; Once more, searchers and other people relevant with search are also paid close attention in holographic search, thereby wish that in search procedure relevant people being searched for promotion obtains finally to improve data; At last, the maximum feature of holographic search is the high praise to people's important value in search, and the realization of three features in front all is to utilize the intellectual resources of people's representative are brought.

Good effect of the present invention is:

The holographic search mechanisms that the present invention proposes is based upon on the new theory and technical foundation, be different from the current internet search system fully, overcome in the current internet search system in many-sided deficiencies such as computational resource, instantaneity, intelligence, cost performance, data volume, granularity, condition managing, intellectual resources, coverage rate, timeliness, personalization, the form of expression, people and interactivity, provide strong technique guarantee for setting up a brand-new search system.

Description of drawings

Fig. 1, current internet search mechanisms;

The operating mechanism of Fig. 2, holographic search.

Embodiment

The first step utilizes the intelligence system service in Internet application system.The essence of computing system difference in the middle of different application is huge.If weigh with the number that intellectual resources participate in, computing system can be divided into two big classes so, i.e. computing system under computing system under the intellectual resources control and the machine algorithm control.The intellectual resources here refer under the conscious situation of the mankind, the intelligent decision of making according to the state of self-demand and current computing environment and judge the corelation behaviour that produces according to this.Corresponding with intellectual resources is exactly machine algorithm.It has been generally acknowledged that machine algorithm can't produce intelligence and can only finish fixing task under finite state according to the order of appointment.And " computing system is under intellectual resources control " refers to use, distribution, the management of computational resource in the computing system and the dynamic change that obtains all to be decided by intellectual resources, but not the machine algorithm of formulating in advance.This produces stark contrast just and with the system that relies on the machine algorithm operation fully.The difference of this two type systematic can be summarized as follows.At first, to the understanding difference of intellectual resources.In the computing system under intellectual resources control, intellectual resources are taken as the important component part of system resource.It is determining the important indicator of computing system except function, i.e. performance and intelligence.To the ignorance of intellectual resources, the function that can cause computing system to provide could guarantee to meet consumers' demand under the situation of computational resource height input.Computing system under machine algorithm control, to the work of intellectual resources in system as any consideration, what think user and computing system in the middle of its potential consciousness is consumption to computational resource alternately, be machine algorithm required that reduce as far as possible with repel; In such system, have only by prior algorithm design just can make full use of computational resource, reduce the pressure that the computational resource excess demand causes system.Secondly, the difference of applied environment.Not all applied environment all is adapted at intellectual resources control operation down.System requirements under intellectual resources control has abundant intellectual resources, and these intellectual resources have the ability of control computational resource.In some applied environments, if intellectual resources are very few or intellectual resources can not exert an influence to the change of computational resource, so such environment just is not suitable for operation under intellectual resources control.The machine algorithm of a maturation is more suitable for such system.If but the system of a suitable intellectual resources control has used the leading mode of machine algorithm, can cause the waste of computational resource or the loss on deficiency and function, performance and the intelligence so.At last, to the utilization difference of intellectual resources.In the computing system of machine algorithm control, computational resource also is under the control of human mind's resource strictly speaking.Algorithm is the embodiment of human mind's resource in computing system in fact.Yet this algorithm is in case formation just no longer changes, and computing system will move according to the pattern of fixing in advance forever.Computational resource is to communicate with each other by the algorithm of prior formulation to cooperate in such system, can finish in high performance environment thereby satisfy basic function.Computing system under the intellectual resources control can't be predicted the computational resource that certain calculating may have in advance.In such system, the number of computational resource is decided by intellectual resources user's judgement in other words; And this judgement is dynamically carried out after system's operation.When the computing system of planing machine algorithm controls, the deviser has a prerequisite understanding, and promptly the computational resource of all systems all is being about under the algorithm controls of formulating, and the computational resource total amount can not change in the whole service process; Even change, also be that the pattern according to prior agreement is adjusted computational resource in the environment of explication.It is small that the computational resource that common this adjustment causes changes, and general all manually the participation under static environment finished.The system of intellectual resources control fully effectively utilizes the computational resource except considering when design, also will consider the mutual variation that causes computational resource owing to user and computing system.Usually such variation is exactly the increase and the minimizing of computational resource.The variation of sample whatsoever, the principle during system design are not to move against such variation, but comply with such variation when guaranteeing fairness as far as possible." comply with " and refer under the guidance of intellectual resources, backup system is set up such open environment.In algorithm design, the change of the computational resource that should cause intellectual resources is monitored; When finding such variation, the method for operation of system must be held and be this change condition that facilitates.

In second step, control Internet application system with intellectual resources.Internet application system is the system that possesses numerous users.Along with the raising of the computing equipment technology and the communication technology, the computational resource that the user had is more and more abundanter, i.e. under user's control, the user can determine these computational resources to participate in or leave certain concrete application to the computational resource of these increases fully.Concerning the internet, applications of a success, its scale does not have the upper limit.This scale is normally weighed with customer volume.This is the typical case of intellectual resources control computing system.Just as discussed earlier,, will have to pay more computational resource if one is suitable for the leading applied environment of intellectual resources and uses that machine algorithm is leading to be calculated, simultaneously can't assurance function, the high-quality of performance and intelligence.Its reason is discussed below.

1) can cause the unreasonable use of computational resource to the ignorance of intellectual resources.When taking as the leading factor with machine algorithm when dealing with internet, applications, because there is fundamental difference in the system of this system and intellectual resources control, the computational resource that causes the user to control brings help not only can for the system that is moving, and can cause the resource request pressure that comes from the user on the contrary.Scale is big more, and then resource pressure is also big more.The pressure of computational resource must cause the loss of application system function, performance and intelligence.This is an aspect of problem.The another one aspect is when ignoring intellectual resources, also may cause the waste of computational resource.Based on the computing system of machine algorithm is system self to provide computational resource to take as the leading factor, and no matter so whether computational resource enough, and system all can only work under such environment.In order to satisfy potential a large number of users demand, can prepare a large amount of computational resources in advance usually.Most of the time because user capture quantity not sufficient and state that machine algorithm can't show such intellectual resources is made a response can cause the waste of computational resource.In computing resource waste or have under the prerequisite more than needed, function, performance and intelligence just have basic assurance.But such assurance is to be based upon on the expensive basis.

2) ignorance of intellectual resources is caused the complexity of algorithm design.Intellectual resources are not merely user's behaviors, and this behavior also causes the number of computational resource in an application.If ignore intellectual resources and relevant computational resource thereof, must cause the complicated of machine algorithm.In current internet was used, the computational resource that must have with the computing system of machine algorithm control was to be in rare relatively state: this is that a large amount of potential users' visits and network capabilities relatively lag behind and cause.Even userbase is at a time relatively little, but network bottleneck all exists at all constantly.Under the rare relatively condition of computational resource, be exactly urgent or even necessary to the utilization of intellectual resources.Otherwise, can only under the limited condition of computational resource, strengthen machine algorithm, make that the utilization of computational resource is more effective.This can cause algorithm too complicated, the cost prohibitive of system design.

3) ignorance to intellectual resources finally always causes simple and crude function, performance and intelligence.In the system that intellectual resources are enriched, use the leading computing system of machine algorithm, finally must cause function, performance and intelligence on a simple and crude basis, to be finished.Even this refers under normal (being that computational resource the is enough) condition of system, also can only provide limited function, performance and intelligence.Between these abilities and real perfect functions, performance and the intelligence very big difference is arranged.The current internet system is exactly an outstanding example.Such system does not appropriately utilize intellectual resources, finally cause such as containing, non-instant, non-stream, stateless, on a small scale, small grain size, low performance, lightweight, low intelligence and problems such as expensive.

In the 3rd step, the distributed computing environment of setting up under the intellectual resources control is the large-scale non-central Internet internet.Since internet system is the computing system that is fit to intellectual resources control, so how set up such system? realize this target in the so special computing environment in internet, concrete steps are as follows.

1) gives the possibility that user's computing equipment participates in calculating.So-called participate in calculating that refer to can the contribution calculation resource to whole computational activity, rather than only appear in the computing system in the mode of consumption calculations resource.According to this definition, in current internet, be actually not as the computing equipment of being controlled by intellectual resources and participate in calculating.The reason that causes this problem is that the current internet system is based on the central distribution formula structure [1] of [1] under the complete thin client/server mode.Complete thin client/server mode has been stopped client and has been participated in the possibility calculated, and promptly client can only be sent computation requests and the waiting for server end is responded.Changing such pattern at first is exactly to make complete thin-client become the client that can share server task; This is changed according to concrete application even can become the client of sharing most of server task.This process can become by the conversion of thin-client to rich client.Secondly, also should consider and utilize collaboration capabilities between the client.A complete thin-client if can also form the model of cooperation with other clients, can make the possibility that participates in mutually between the internet computing equipment calculating reach the highest after being transformed into rich client so.At this time Zui Da change should be to make a client except the ability that possesses the request of sending, and also possesses the ability that accepts request and handle request.This is the prerequisite that forms cooperation between client.After a client possesses request, is requested, handles request and the ability that gives a response, just be transformed into peer client by a client.Certainly, the possibility that just participates in calculating of computing equipment is given in so simple transformation, also has gap from perfect peer client.At last, a way that more possesses universal significance is that server is become peer server.This variation can make all computational resources all possess the possibility that participates in a calculating on computation schema.

2) give the power to make decision of user's computational activity.Possess the possibility of the calculating of participating in when computing equipment after, and do not mean that they are in the optimum condition that just can reach under the control of machine algorithm on function, performance and the intelligence.Tracing it to its cause is because machine algorithm does not possess real intelligence, can not make the judgement of meeting consumers' demand for the user in complicated and diversified computing environment under the instant or even overtime situation.Just because of the deficiency of machine algorithm, must incorporate user's intellectual resources in the middle of the calculating.It is not difficult to allow the user obtain such power to make decision, and it is much simple to produce the such target of intelligence than machine algorithm at least.Usually can just can reach this purpose by succinct interface, possible result after allowing the user clearly understand the possible selection of ongoing calculating in the middle of the system, correlation behavior, user and user to make a choice.When these information were presented on the interface, the user just can rely on the intelligence of self to make it to think suitable selection, and perhaps current at one's discretion calculating is proceeded, and perhaps selects more interested calculating, perhaps interrupts all calculating.The computational activity right to choose is to incorporate intellectual resources in the middle of the internet, applications and finally obtain the intelligence service to have set up prerequisite.

3) utilize machine algorithm to remedy the deficiency of intellectual resources.Intellectual resources and do not mean that machine algorithm is powerless in system after having possessed the right to choose of calculating.In fact, outstanding machine algorithm should combine with appropriate use intellectual resources and just can remedy not enough each other and obtain optimum computing system.Though intellectual resources possess the unexistent absolute predominance of machine algorithm on intelligent decision, but because the user makes judgement according to the hobby of self, this might not be best to the overall situation at any time, even some decision-making also is harmful to user itself; Simultaneously, intellectual resources are compared with machine also weak point, as pure numerical evaluation ability, do not possess statistical power of global state or the like.At this time need computing system the user to be given a hand by appropriate mode.

4) set up the elemental network topological structure that intellectual resources drive.After conventional client and server have been transformed into peer client and peer server, in fact also make current internet change the basic topological structure that intellectual resources drive into by division center.This also is a key means of realizing that intellectual resources effectively incorporate.This topological structure builds on the understanding of user self to computing system fully, be the judgement adding current ongoing calculating in internet that the user can arbitrarily pass through oneself, and in the reciprocal process of the computing equipment of its control and computing system, constitute the new basic topological structure in internet.This basic topological structure is to be based upon on the basis that user self judges, itself is exactly the network structure of intellectual resources of having condensed.This structure is that the basis that possesses the new function of the high intelligence of high-performance is provided in whole internet system.Certainly, just as mentioned earlier, the user freely judges because self reason is not always correct.At this time, computing system can provide possible selection (as strong node, by force use and option such as main activities) assisted user enters system, makes new network topology structure more accurate to the assurance of intellectual resources.More this topological structure itself do not have rule with division center, actual discovering because the employing of aforesaid way, the computing equipment of final a large number of users control can form the network model that possesses unique rule, and promptly community network [6] is called complex network [9] again.Based on this network, computational resource obtain and cooperative association becomes more reasonable under the guide of intellectual resources.

5) the suitable routing algorithm of imparting system.In distributed system, routing algorithm is the core algorithm of system.The essence of routing algorithm is to obtain the method for computational resource.In the current internet system, because computed altitude is controlled under the machine algorithm, its mode of obtaining resource is also finished by machine algorithm.General idea all is the fast processing that thereby the computational resource absolute ruleization is fit to machine algorithm, such as center registration law [1] or regular texture method [10].The former is fit to the distributed computing system based on division center.Current internet is sought computational resource with regard to being based on such structure.The latter is fit to the large-scale distributed computing system of non-headed construction.This computing environment also is a community network or complex networks system in essence.But, the routing mode of this rule-based structure is because the intellectual resources of not utilizing community network itself to possess, plan Internet resources in the mode of machine algorithm purely, finally safeguard the burden of this dummy rules topological structure even surpassed the repayment that obtains in the searching resource.The routing algorithm of a suitable community network feature must be considered the feature of complex network simultaneously and be based upon the concrete characteristics of using on this network topology structure.

6) the suitable swarm algorithm of imparting system.In order to set up the distributed system that intellectual resources are taken as the leading factor, the problem that another one need be considered is exactly the method for after obtaining possible computational resource by routing algorithm these resource organizations being got up.The network foundation of internet is the system of a dynamic change, and as also being in the middle of the change by the computing equipment of user control or by the directly actuated computational resource of intellectual resources on this system.This mainly is that to come from the computational resource that computing equipment has relatively limited, especially is subjected to the especially severe that influences of unstable networks.In addition, the selection of user in computation process also can influence the stability of calculating.So when organizing such computational resource, will possess two condition precedents: the first, calculating itself will possess the meaning that the user can experience; The second, it is enough big that the scale of computational resource is wanted.Between these two conditions is associated.Have only the user to experience the meaning of calculating in computation process, the scale that so just might make enlarges; And the expansion of scale is the computational resource that significant calculating attracts more users control.After these two condition precedents satisfied, these computational resources will be organized into the collaborative work cluster according to the principle of reciprocity.When above-mentioned two conditions did not possess, the large-scale non-central Internet internet can provide the most basic performance to guarantee.Performance at this time can descend seriously, prevents cluster resource scarcity on a small scale unless the large-scale non-central Internet internet providers has abundant computational resource.

In the 4th step, reuse the current internet resource under the non-intellectual resources control.After having set up the large-scale non-central Internet internet, be not to have set up the system that opposes with current internet.Though the two working mechanism's difference, new system can merge existing system.Have a large number of users, data and Service Source in the current internet.Take suitable means that these resources are incorporated in the new internet environment, can bring providing powerful support for of intellectual resources and computational resource for this new system.Basic skills be by the periodic access website using obtaining its data or service, and multicast in the topological structure of intellectual resources control.By utilizing these resources, can progressively advance the large-scale non-central Internet Internet development, realize that finally all internet computing equipments all move in new ways, realize more perfect internet system.

In the 5th step, participate in guaranteeing enough search service computational resources by all computational resources.Holographic search is based upon on the large-scale non-central Internet internet, why can possess the feature of " holography ", exactly because there is relative sufficient calculation resource to guarantee.For search, at first be data resource.Because all computing equipments that add the large-scale non-central Internet internet system all have the possibility of the calculating of participating in, so it is ready that the data of sharing will become the part of total system.Such data resource need not to do centralized moving on the internet by the way of " reptile ", has saved extra computational resource and has enlarged data coverage.Secondly, cpu resource also can be guaranteed on this platform.Owing to taked high dynamic cluster technology, in fact the pressure of search can not concentrate on the minority super node on the complex network; Searching request is responded by whole dynamic cluster by the method for balancing dynamic load.Once more, storage resources also is an important computations resource of supporting search.Storage resources can be divided into lasting storage resources and instantaneous storage resources.The former is because system allows any equipment to enter and participates in calculating, and the lasting storage resources of numerous equipment itself just becomes the part of system.This has greatly enriched the total amount of storage resources.The latter generally is used for preserving dynamic-change information in the search procedure, as the computing equipment current state.Such storage resources can come from dynamic cluster equally, and its principle is also simple relatively, i.e. the cluster that the computing equipment of these users control forms in a certain amount of time has similar demand, thereby its state also has similarity.This has significantly reduced state and has preserved the instantaneous space consumption that need pay.It is emphasized that these two kinds of storage resources do not do strict the qualification on storage medium, promptly lasting storage device also can be preserved as instantaneous state, but is not all right conversely.At last, bandwidth resources are frequent and to provide the search service of heavyweight also be important concerning searching request.Based on the use of high dynamic cluster, the bandwidth pressure to the search data source in the transmission has also imputed to the bandwidth that computing equipment had in the cluster equally.

In the 6th step, under the intellectual resources guiding, search for.The solution of this problem comprises concrete steps.

1) the searched data of distribution under the control of intellectual resources request.Initiatively internet data to be copied to the computational resource of self control different with current internet, and the data in the large-scale non-central Internet internet system need not such duplicating, because data wherein have the distribution characteristics [9] of self uniqueness.This distribution characteristics is embodied in aspects such as physical location, geographic position, backup number, machine capability even crowd's background.Physical location on the internet is to reduce the key factor of network bottleneck to the system effectiveness influence.When the access internet data, be normally subject to the physical location on the network, it is many that the backup of accessed data also can present the data that the near computing equipment of physical location and data source has; And the data that physical location and data source computing equipment far away has are few.On the internet, the physical location of residing geographic position of computing equipment and computing equipment might overlap, and also may fall far short.In general data all possess regionality, so the distribution of data on diverse geographic location also has its rule.The obvious difference of backup number in the large-scale non-central Internet internet.Satisfy its backup degree height of data of user's needs; Otherwise, then low.Because the user can control computing system, has backup also can be left out in user's maintenance process even can not satisfy the data of user's needs so.Usually, the data that keep in the large-scale non-central Internet internet all are the data of being accepted by the user on the whole, and the backup number is directly proportional with the data pouplarity.In addition, DATA DISTRIBUTION also can show difference because of the difference of machine capability.For the abundant machine of computational resource, data that it has and data character all can have difference with the machine and the data character of computing power difference.Such as, the machine that computing power is strong may have more data, and may comprise more heavyweight data etc. in the data.At last, different crowd's backgrounds also can impact the data distribution.The data homogeneity degree height that the computing equipment of same people background can have, and the data degree of homogenization that different background crowd's computing equipment has is low.Need emphatically point out be these features can the large-scale non-central Internet internet based on complex network on embody, can embody intellectual resources control search system is significantly different with current internet systematic search system down so depend on complex network search.Yet this different embodiment needs the auxiliary of machine algorithm and to the assurance of user's dynamic interaction, relies on complex network or conditional merely.

2) on complex network, spread searching request, make it processed in the system that meets the intellectual resources regularity of distribution.Will be at the enterprising line search of complex network, what at first will solve is exactly the problem of searching request diffusion.Different with the current internet search mechanisms, the target that searching request sends is uncertain at complex network: if there is a center search peer server, so just can equally with current system be sent to this server; If there is no, then need seek in the node of neighborhood and determine having with this node or equipment.Neighbours' selection is very important to the search efficiency in the complex network.In the large-scale non-central Internet internet, should determine neighbours according to the interaction feature of concrete application.Neighbours can be understood as and enter the concrete inlet of using in large-scale non-central Internet internet.This inlet is selected correct, can bring important support in this application of whole participation or calculating for the user.Neighbours determine what the accumulation of user's intellectual resources in concrete calculating or application brought.After neighbours determine, thereby just can select searching request is spread and then to its processing to neighbours.This selection course is according to carrying out with neighbours' intersection record in the past.Most possible Data Source be can judge by this record, this or relevant a plurality of neighbor nodes then request issued.These nodes can be searched on the equipment of self controlling after obtaining request, if find, then can return; If come to nothing then can send to the node adjacent to searching request according to similar principle with it.From top process as can be seen, searching request is not blindly to spread in the large-scale non-central Internet internet, final such diffusion be under the prerequisite that meets the intellectual resources regularity of distribution, carry out and with very short speed [11] near data source.

3) handle searching request in the computational resource under appropriate intellectual resources control.Because the diffusion of searching request finally can cause searching request to arrive data source with two-forty in the large-scale non-central Internet internet hunt, so searching request can be handled by appropriate computational resource.This computational resource may be single computing equipment, and is a high dynamic cluster under more situation.This cluster is because assisting of intellectual resources generally can possess one or more features of DATA DISTRIBUTION noted earlier.

4) basis and intellectual resources reciprocation are set up the relation of Search Results.If searching request all is to express in the so fuzzy mode of key word usually.This mode is the compromise result between human's demand expression and machine algorithm feature.The human's demand abnormal expression is abundant, as literal, voice, expression and gesture or the like.But the expression of these demands has only few part to be far from being understanding certainly by machine processing.Key word is exactly wherein a kind of.Because the restriction of expression way and machine algorithm should be the process that highly intelligentized search procedure has been degenerated to keyword matching.Such process can't cause correct result's generation, has numerous matching results usually and occurs; And only be that the result who comprises key word occurs, and result relevant with key word but that do not comprise key word can not occur.In the face of such problem, the current search system has to Search Results is sorted.At least the thought of the ordering thinking of current internet search service and the holographic search part that communicates promptly all utilizes human mind's resource to carry out, as the PageRank technology.But because available intellectual resources are limited and computational resource is also limited in the current internet, so cause only having in the search procedure ordering to consider intellectual resources.In holography search, available intellectual resources are abundant, and all condense on the complex network of dynamic change, and this has set up the basis in the search Search Results further being differentiated.Because, have special " a watching " function and mutual strong feature in the large-scale non-central Internet internet, the system that can assist search after the relation that analyzes between Search Results, still can with user's reciprocal process in carry out dynamically adjusting immediately.

In the 7th step, realize holographic principal character of searching for.Discuss the technical foundation of holography search the front.This section will be discussed to the implementation that essential characteristic is searched in holography, and its relevant concrete steps are as follows.

1) utilize all Internet resources to reduce searching cost.Search service is provided, and especially parallel search service needs huge cost input to handle, store and visit the calculating pressure that brings to deal with mass data.This input demand to duplicating the degree of whole internet, almost has been the stage that nobody can bear except only a few enterprise.Thorough change has taken place in the foundation structure that holographic search system is relied on, and main computational resource comes from and participates in calculating computational resource that the user controlled and to the reusing of current internet system resource, so the input of computational resource is much smaller.Simultaneously, the supplier of holographic search can also provide an amount of computational resource to come the node of inadequate resource in the backup system.Have Zipf Law[12 because the internet calculates] feature and social property [6], in the middle of causing most of computational resource can be put to that minority is crucial calculating.After these resources are provided by the computational resource of user control, can between the resource of user-driven, constitute cluster, this cluster can replace the computational resource of searching for the supplier, becomes the resource that search mainly is provided.Like this, search supplier's resource becomes the auxiliary resources in the search service: the cost of search service reduces greatly.

2) pass through the instantaneity that strong flexible cluster is realized search system.The current internet system is owing to must copy to this locality to the data that are on the Internet Server, and in the face of the huge internet system, this calculated amount is huge, can't guarantee to obtain immediately the data variation on the Internet Server.In order to address this problem, the mode that the current search system takes to monitor the main website of minority remedies.But this mode does not fundamentally change current internet search system fault of construction on instant problem.In holographic search technique, any one computing node all and formed the cluster of strong retractility between all the other nodes.The variation of one of them node data can effectively be spread in this cluster and is unlikely to cause system pressure or obstruction.Certainly the prerequisite that reaches this purpose be in the social environment that complex network constitutes any to create back end constantly be minority; A cluster is as in the main modular that makes up complex network similar situation being arranged also.Reality also is like this.Otherwise, when all in the network system or great deal of nodes all have data variation and when whole cluster even whole complex network disseminated, the strong retractility of system also can be destroyed.In order in search, to realize instantaneity, be exactly to form high dynamic cluster between search requestor and the data source in essence.Any variation in the cluster can be embodied in all nodes in cluster, and instantaneity has just guaranteed like this.

3) the intelligent resource of utilizing system itself to be comprised realizes strong intelligence.For the generation of intelligent result in the search, has discussed much the front; Can be referring to the argumentation of " first step " and " second step ".Its cardinal rule can simply reduce and utilize intellectual resources, uses in conjunction with concrete simultaneously, come fully data and in the past, current intellectual resources combine, thereby the intellectual resources of existence are showed to the user by computing system.

4) by utilizing whole computational resources and dropping into limited resources and realize the search high-performance.Here the high-performance of mentioning is high performance-price ratio strictly speaking, and the new feature that promptly holographic search has is not a prerequisite with more substantial computational resource input.Reason is that the foundation of large-scale non-central Internet internet environment causes all computing equipments can participate in correlation computations; And the adding of these computing equipments has alleviated search service supplier's cost greatly.When cost reduces, computational resource can increase greatly on the contrary and the combination of resource also more reasonable.These advantages all make based on the advantage of search service on performance of this platform bigger.Certainly, a special requirement is arranged in search service, i.e. response is wanted rapidly.Even Search Results quality height, if but after the user sends searching request, can not in finite time, give a response, also can hinder search system to be accepted.After all computing equipments all participate in searching and computing, also can exist searching request in system, to disseminate and cause the problem of low-response.For addressing this problem, can suitably increase computational resource in system.Usually this computational resource can be arranged according to the geographic position, preserves all data directories in this geographic range, and with this zone in all computing equipments form search service clusters, to improve search response efficient.

5) multicast of supporting by cluster realizes the search towards the heavyweight data.After obtaining Search Results, system can transfer possible computational resource and with the form that flows data directly are transferred to search requestor under the multicast protocol that with reciprocal is principle be supported.Even comprise heavyweight data (as video etc.) in such data, holographic search service also can be handled.

6) utilize intellectual resources to organize Search Results to form coarsegrain.The coarsegrain that guarantees search is actually the logic integrality that guarantees Search Results.In search procedure, search service can according to intellectual resources to the relation between Search Results (such as ordering) determine.This relation can be linear, also can the level type.After search data forms relation, just have the granularity problem of corresponding base in the Search Results node.In the holography search, under the assurance of heavyweight host-host protocol, the data of each node can be transmitted afterwards according to the elder generation of its relation.

7) utilize the equipment of search requestor control to realize having state search.In fact the consumption to resource is very big in the state maintenance.The large-scale non-central Internet internet has and has advantage in this.But it is to be noted and realize that it is most of node state advolutions in complex network that the prerequisite of state search is arranged, and the minority node possesses the ability that produces raw data.These minority nodes cause the state that occurs in the system under the various demand correspondence just.From the overall situation of whole complex network, it is very little that the shared resource of various states is compared with the total system scale.Simultaneously, various states have formed dynamic cluster round the minority super node, are convergent at a cluster internal state promptly.So just solved the consumption problem of state maintenance to resource.Exist the node that quantity does not wait a cluster inside, these nodes obtain data or service from super node jointly, and cooperate, bear state pressure each other and improve performance between them.

8) allow all devices to add search and realize high coverage rate.Because in the search service based on the large-scale non-central Internet internet, any computing equipment can become a member of searching and computing, in fact just farthest guaranteed the coverage rate of search to data.This has been avoided used device in the current search system, duplicate and bring in constant renewal in the internet data that is replicated causes the low problem of coverage rate.We can say that the acquisition of this high coverage rate is the result who respects the original thought in internet, promptly any computing equipment all should become the node of par on the internet, have to participate in the right that the internet calculates, rather than the minority node could participate in internet calculating.

9) under the support of instantaneity, realize possessing the search of temporal characteristics.Temporal characteristics is combined closely with instantaneity.If a search service system can not obtain data immediately, also can not be to defining that the temporal characteristics of data is unified.Description by the data time itself is in the indemnifying measure of losing after obtaining data capability immediately fully.But because can not be to holding the immediate status of searched resource, in fact quality evaluation in a certain amount of time breaks away from data to cause static time description.In the search system of multiplexing number evaluating data quality standard, promptly allow to obtain the publisher time of data is described with hyperlink current, describe but be difficult to obtain the time that the corresponding hyperlink of these data is re-used.In fact such search system can't carry out temporal measurement to data.Have only in other words long term data is had evaluation capacity.So-called long term data refers to has issued the long enough time on the internet, and the hyperlink of this data correspondence number of times that is re-used is enough to become the standard of weighing its quality.Time is one of standard that embodies data value.The data assessment of current this single time form can not satisfy the user to the requirement on the data timeliness.After the user sends request, the data that necessarily require search service to provide to satisfy the demands just.Because in the process that obtains internet data, except hyperlink is multiplexing, do not have effective ways to weigh the value of data, cause and to present to the user to the data after the long-term assessment (data of the maximum multiplexing correspondence of hyperlink) at first; On the other hand, the keyword matching result is numerous usually, and the user is out of patience to differentiate from a large amount of return results, and only can obtain possible outcome from the data that come the front.This has caused effectively utilizing the consequence of passing judgment on the quality of data alternately with the user.Like this, the multiplexing insensitive mode of this time of hyperlink has just become the current search technology to weigh unique effective standard of the quality of data.In this case, the data that search system is preserved are many more, and the data possibility of flooding user's request is just big more.This is that current internet does not possess one of problem that the immediate assessment data capability brings.Because holographic search system possesses the ability of obtaining data and variation thereof immediately, the while has multiple by utilizing the strategy of intellectual resources assessment data again, and therefore such search possesses timeliness.

10) utilize based on the non-machine algorithm realization of intellectual resources personalized search.Realize that by machine algorithm personalization all is being proved to be unsuccessful qualitatively with on the efficient in the internet hunt field.The personalized method cardinal rule of holographic search technique realization is still utilized the intellectual resources in the system.Realize that personalized intellectual resources embody a concentrated reflection of on a plurality of high dynamic clusters in the large-scale non-central Internet internet and the computing equipment that each is concrete.These computational resources under the control of user's intellectual resources have truly reflected the individual character of controlling its user just.After a user sent searching request, in fact different with the current search service, its search can launch in the overall situation, also can launch in a cluster, and finally actual is to carry out on a computing equipment.Because computing equipment has shown the individual character of controlling its user, thereby causes same searching request can obtain different results in different hunting zones.The hunting zone can be specified by the searchers, also can judge according to state of user.Because holographic search is a kind of state search that has, this has set up prerequisite for personalized search.

11) realize the multi-format data search based on heavyweight processing power and multi-format data combination.The multi-format data search is to be based upon on the large-scale non-central Internet internet that possesses the heavyweight data-handling capacity.Possess the ability of processing heavyweight data when a platform after, user wherein just has been endowed the freedom of issuing various formatted datas.Usually these multiple forms can effectively be made up and more clearly be expressed user's intention, as the combination of literal, picture, audio or video.When the data of this combination appear in the system, also just set up at the search of such data.Also have a problem to be in the native system, when the current internet resource is transformed, because the shortcoming of current internet ability, the data of these conversions still present its original looks in the large-scale non-central Internet internet, native system does not make up by the data that machine algorithm gives these different-formats.

12) based on the search of community network realization towards the people.The internet final purpose is to set up interpersonal interchange, rather than the exchanging of simple people and machine or people and data.In the holographic search system, by data resource in the Search Results can be direct with relevant the intellectual resources correspondence.Because general search all is based on high dynamic cluster, so also can be found out based on the crowd of this high dynamic cluster.In fact, can also carry out based on ordering these relevant individuals, crowd, to improve final search effect the people to system's contribution amount.

13) utilize multicast system to realize interactive searching.The intelligence that powerful again computing system also can't replace the people to have.After Search Results comprised people and crowd, the user can carry out alternately with these people, crowd and relevant computational resource, thereby obtains better Search Results, remedies the deficiency of search technique.In the large-scale non-central Internet internet, mutual cost is born by all computing equipments that participate in to calculate, make in the search alternately not only can be based on the lightweight data, also can be based on the heavyweight data; Not only can be based on one to one, also can be based on many people.But it is pointed out that when based on one to one or during few mutual of participant,, need the search supplier to arrange that computational resource assists because computational resource is limited.This is the weakness of large-scale non-central Internet internet on alternately.But after being aided with the extra computation resource, its effect can not be lower than the current internet interactive system.

14) comprehensive previous methods finally forms holographic search system." holography " can have the understanding of two aspects, and the one, the embodiment of intelligence in the search, two are meant and can obtain multiple different Search Results.The cardinal rule that realizes these targets is exactly the reasonable utilization to intellectual resources and computational resource that intellectual resources are controlled.Comprehensive various features noted earlier and its implementation have just realized high intelligence, high-performance, low cost, omnibearing search system, promptly holographic search system.

List of references:

【1】Jean?Dollimore，Tim?Kindberg，George?Coulouris；Distributed?Systems：Concepts?and?Design；Addison?Wesley，4th?Edition，2005，ISBN：0321263545

【2】Hari?Prasad?Thadakamalla；Decentralized?Algorithms?for?Search?and?Routing?in?Large-Scale?Networks，PhD?Dissertation，Pennsylvania?State?University，December?2007

【3】Semantic?Searching；http://www.powerset.com

【4】Luger，George；Stubblefield，William；Artificial?Intelligence：Structures?and?Strategies?for?Complex?Problem?Solving(5th?ed.)，The?Benjamin/Cummings?Publishing?Company，Inc.，ISBN?0-8053-4780-1

【5】Sergey?Brin，Lawrence?Page；The?Anatomy?of?a?Large-Scale?Hypertextual?Web?Search?Engine，Proceedings?of?the?Seventh?International?Conference?on?World?Wide?Web?7，Page(s)：107-117

【6】Yilei?Shao；Exploring?Social?Networks?in?Computer?Systems，PhD?Dissertation，Princeton?University，June?2007

【7】Google；http://www.google.com

【8】Ying?Yu；Measuring?and?Modeling?the?Web，PhD?Dissertation，Stanford?University，July2008

【9】Newman?M?E?J.；The?Structure?and?Function?of?Complex?Networks，SIAM?Review，2003，45，Page(s)：167-256

【10】Rowstron?A.，Druschel?P.；Pastry：Scalable，Decentralized?Object?Location?and?Routing?for?Large-Scale?Peer-to-Peer?Systems，in?Proceedings?of?the?2001ACM?SIGCOMMConference?on?Applications，Technologies，Architectures，and?Protocols?for?Computer?Communication(SIGCOMM’01)，San?Diego，CA，August?2001，page(s)：247-254

【11】Hari?Prasad?Thadakamalla；Decentralized?Algorithms?for?Search?and?Routing?in?Large-Scale?Networks，PhD?Dissertation，Pennsylvania?State?University，December?2007

【12】George?K.Zipf；Human?Behavior?and?the?Principle?of?Least-Effort，Addison-Wesley，1949

Claims (8)

1. a method of setting up the internet holographic search service is characterized in that, described holographic search service method comprises utilizes intellectual resources to serve Internet application system; Control Internet application system with intellectual resources; The distributed computing environment of setting up under the intellectual resources control is the large-scale non-central Internet internet; Reuse the current internet resource under the non-intellectual resources control; Participate in guaranteeing enough search service computational resources by all computational resources; Under the intellectual resources guiding, search for; Realize the principal character of holographic search.

2. a kind of method of setting up the internet holographic search service as claimed in claim 1 is characterized in that the described intellectual resources of utilizing are served Internet application system; The specific implementation step is as follows:

1) intellectual resources is used as the important component part of system resource, by the important indicator of its decision computing system except function, i.e. performance and intelligence;

2) in the Internet application system that intellectual resources are enriched, make full use of intellectual resources, to improve internet function, performance and intelligence;

3) the variation of algorithm operation back user and system generation as Consideration, design appropriate method and utilize intellectual resources.

3. a kind of method of setting up the internet holographic search service as claimed in claim 1 and as claimed in claim 2ly in the abundant Internet application system of intellectual resources, make full use of intellectual resources, it is characterized in that, describedly control Internet application system with intellectual resources; The specific implementation step is as follows:

1) pay attention to using intellectual resources to guarantee the reasonable use of computational resource;

2) utilize the design of intellectual resources shortcut calculation.

4. a kind of method of setting up the internet holographic search service as claimed in claim 1 is characterized in that, the described distributed computing environment of setting up under the intellectual resources control is the large-scale non-central Internet internet, with this calculating basis as the holography search; Its specific implementation step is as follows:

1) gives the possibility that user's computing equipment participates in calculating; Make complete thin-client become the client that to share server task, utilize collaboration capabilities between the client, make it to become peer client; And the transformation common server is a peer server;

2) give the power to make decision of user's computational activity, the user can rely on the intelligence of self make its any suitable selection;

3) utilize machine algorithm to remedy the deficiency of intellectual resources;

4) set up the elemental network topological structure that intellectual resources drive; The judgement that the user can arbitrarily pass through oneself adds the current ongoing calculating in internet; And in the reciprocal process of the computing equipment of its control and computing system, constitute the new basic topological structure in internet;

5) the suitable routing algorithm of imparting system, help system is obtained computational resource;

6) the suitable swarm algorithm of imparting system after obtaining possible computational resource by routing algorithm, gets up these resource organizations.

5. a kind of method of setting up the internet holographic search service as claimed in claim 1 is characterized in that, described current internet resource of reusing under the non-intellectual resources control; Its specific implementation step is as follows:

1) by the periodic access website using to obtain its data and service;

2) multicast in the topological structure of intellectual resources control.

6. a kind of method of setting up the internet holographic search service as claimed in claim 1 is characterized in that, and is described by the enough search service computational resources of all computational resources participation assurances, thereby guarantees for the holography search provides sufficient resources; Its specific implementation step is as follows:

1) utilize the user to be ready the data shared;

2) take high dynamic cluster technology to utilize cpu resource;

3) by allowing any equipment to enter and participating in calculating, utilize its storage resources;

4) based on high dynamic cluster, marry again bandwidth pressure.

7. a kind of method of setting up the internet holographic search service as claimed in claim 1 is characterized in that, describedly searches under intellectual resources guiding; Its specific implementation step is as follows:

1) the searched data of distribution under the control of intellectual resources request; Utilize the distribution characteristics of data self uniqueness, comprise aspects such as physical location, geographic position, backup number, machine capability and crowd's background; The use machine algorithm is auxiliary, and holds user's dynamic interaction;

2) on complex network, spread searching request, make it processed in the system that meets the intellectual resources regularity of distribution; Interaction feature according to concrete application is determined neighbours; According to neighbours being selected, thereby searching request is spread and then to its processing with neighbours' intersection record in the past; Neighbor node is searched on equipment of itself; If find, then return; Otherwise, searching request is sent to the node adjacent with it according to similar principle;

3) handle searching request in the computational resource under appropriate intellectual resources control; Can on single computing equipment, handle searching request, also can in high dynamic cluster, handle searching request;

4) basis and intellectual resources reciprocation are set up the relation of Search Results; With the key word is the main searching request of sending; Bear results by keyword matching, and utilize intellectual resources to sort; Complex network by dynamic change is further differentiated Search Results.

8. a kind of method of setting up the internet holographic search service as claimed in claim 1 is characterized in that, the described principal character that realizes holographic search; Its specific implementation step is as follows:

1) utilize all Internet resources to reduce searching cost;

2) pass through the instantaneity that strong flexible cluster is realized search system;

3) the intelligent resource of utilizing system itself to be comprised realizes strong intelligence;

4) by utilizing whole computational resources and dropping into limited resources and realize the search high-performance;

5) multicast of supporting by cluster realizes the search towards the heavyweight data;

6) utilize intellectual resources to organize Search Results to form coarsegrain;

7) utilize the equipment of search requestor control to realize having state search;

8) allow all devices to add search and realize high coverage rate;

9) under the support of instantaneity, realize possessing the search of temporal characteristics;

10) utilize based on the non-machine algorithm realization of intellectual resources personalized search;

11) realize the multi-format data search based on heavyweight processing power and multi-format data combination;

12) based on the search of community network realization towards the people;

13) utilize multicast system to realize interactive searching;

14) method of comprehensive front finally realizes holographic search system.

Images