WO2002042934A1

WO2002042934A1 - Method and system for managing data modelled in the form of objects, and method and system for constructing and managing pages using said method

Info

Publication number: WO2002042934A1
Application number: PCT/FR2000/003312
Authority: WO
Inventors: Randal Walker
Original assignee: Randal Walker
Priority date: 2000-11-27
Filing date: 2000-11-27
Publication date: 2002-05-30
Also published as: AU2001218703A1

Abstract

The invention concerns a method for managing data in the form of objects organised into classified objects, each object comprising identification data and a list of attributes. Said method comprises: storing objects <i>in an object storage system</i> <i>including serialization of object files such as XML</i> format files; indexing the attributes associated with the stored objects, to constitute an index in an indexing system including a relational database; an access, within said indexing system, to object identification data in response to requests concerning the attributes indexed therein; and an access to objects thus stored in response to requests concerning their identification data, said stored and accessed objects and the attributes associated with said objects passing through a cache whereto are linked the object storage system and the indexing system. The invention is in particular useful for producing and managing connecting Web sites, for example, in the field of fashion.

Description

"Method and system for managing information modeled in the form of objects, and method and system for constructing and dynamic management of pages implementing this method" "

The present invention relates to a method for managing information modeled in the form of objects. It also relates to an information management system, as well as a process and a dynamic construction and management system implementing this process.

Object databases are widely used because it is possible to contain within each element a multitude of codes and attributes which evolve without requiring a restructuring of the database. The object database is accessed via a unique reference such as the name of the file in the case of a patient in a hospital. However, relational searches in an object database prove to be slow or almost impossible. Conversely, relational databases allow rapid access to well-structured data but cannot deal with elements that change over time. Adding new attributes requires restructuring the database and / or the database indexes. Some databases such as Oracle databases tend to simulate objects, but access remains slow. To remedy this drawback, there is a tendency to use more and more powerful computers, with the consequence of an increase in hardware and software costs.

There are several models for structuring information with the aim of constituting databases, such as object-oriented models, relational models, or hierarchical models. The object model favors the navigation task while the relational model favors the research task, a hybrid solution combining these two models can be considered.

In the management method according to the invention, the storage function is entrusted on the one hand to an object storage system for example in the form of an object base from which only known objects are requested from their identifier, and on the other hand, to an indexing system in the form of a relational base which provides object ID identifiers in response to queries on the attributes which are indexed there.

This therefore amounts to storing objects in their entirety, and to dealing in parallel with the problem of indexing attributes. This indexing can be optimized according to the most commonly considered searches, and observed, by the construction and dynamic management of directories and dictionaries specific to the main requests. The independence of indexing with regard to storage introduces strong redundancy, but offers better security by the possibility of reconstructing a corrupted index from the data already recorded, with reference to FIG. 13. It also makes it possible to adapt the size and structure of the indexes at all times, and to optimize the system for its actual observed use, and not from a purely theoretical conception. Relationships are expressed in objects in memory using properties or attributes, and are treated as such. However, access to certain objects in isolation is not relevant. Larger objects can be stored independently if their content is not always essential for the use of the object that references them. For example, the participation of an individual will be saved in a container which is loaded into memory only on explicit request.

Thus, the document WO9909494 discloses a software methodology allowing a coexistence between a relational database and an object database. The document US5212787 discloses a method and system for accessing a relational database without leaving an object-oriented environment. In this method, a requesting object located in an object-oriented environment sends a data access message to request data located in a relational database outside the object-oriented environment. This message is sent to an interface that creates a data object containing instance variables to carry query results. This interface executes a data access routine and the results are placed in the data object and sent to the requesting object by sending a pointer to the data object.

Document US5819086 discloses a computer system for creating semantic object models from existing relational database schemas. This system analyzes catalog information from the relational database schema and creates a semantic object for each table defined in the catalog. For each column defined within a table, a simple value attribute is added to the semantic object created for this table. The system then analyzes the relationship information stored in the catalog to create object link attributes that define relationships between two or more semantic objects, as well as to create multivalued group attributes and multivalued simple value attributes. If the data catalog does not include relational information, the user is asked to indicate associated semantic objects. The documents US5615362 and US5706506 disclose a method and system for managing relational data in an object cache. The system includes a processor used to construct a plurality of object instances each having their own unique object identifier ID which provides addressing between the object instance and at least one row in the structured database. The processor is also used to build a single cohesive data structure, called an object cache, which includes all object instances and which represents the information retrieved from the structured database in a form suitable for use by one or more oriented applications. -object. An object cache management mechanism implemented in this system has three properties: one concerns the implementation of a "key swizzling" technique using relational pointers between object instances in the object cache to reduce the volume of requests addressed to the structured database. A second property is that only one copy of an object instance is in the cache at a given time even if several different queries return the same information from the database. Finally, this mechanism is presented as guaranteeing the integrity of the data in the cache by appropriately locking the data in the structured database during a database transaction, by unloading the cache data at the end of each transaction, and transparently re-reading the data and re-acquiring the appropriate locks for an object instance from which the data has been unloaded. Document US5542078 discloses an object-oriented data storage integration environment for the integration of object-oriented databases and non-object-oriented data structures. This environment implements a distributed object database application and an object database management system. This application can manipulate data in data stores that include foreign data that is addressed and converted to objects for use by object applications. A storage management API application program interface provides a functional interface for processing objects, referencing them, iterating, indexing object transactions, and managing the cache. This storage management management API is part of a modular architecture which includes an external storage management system which implements classes providing the foundations for engaging external data storage, and which addresses and converts data external objects that can be manipulated by an application using the object database management system.

US6012126 discloses a system and method for hiding non-uniform size objects using partitions of multiple LRU stacks in a range of dimensions. This method implements a cache logic comprising a selection logic and an admission control logic which determines whether an object not currently in the cache can be hidden. This admission control logic uses an auxiliary LRU stack which contains the identities and timestamps of objects that have accessed recently. In this way, the memory required is relatively small. The auxiliary cache serves as a dynamic popularity list and an object can be admitted to the cache if and only if it appears on the popularity list. The selection logic selects one or more of the objects in the cache which must be purged when a new object enters the cache. The order of removal of the objects is treated according to priorities based on both the size and the frequency of access of the object and can be adjusted according to a time to obsolescence factor (Time To Obsolescence: TTO). In order to reduce the time required to compare the space-time product of each object in the cache, the objects can be classified into ranges according to geometrically increasing intervals.

The aim of the present invention is to propose a new concept for managing information modeled in the form of objects organized into object classes, which makes it possible to combine object-oriented models and relational models in an even more efficient way. than with the methods of the prior art.

According to the invention, the method for managing information in the form of objects organized into object classes each comprising identification data and a list of attributes, is characterized in that it comprises:

- storage of objects in an object storage system including serialization of object files such as files in XML format,

- an indexing of the attributes associated with said stored objects, to constitute an index in an indexing system including a relational database,

- access, within said indexing system, to object identification data in response to requests for attributes which are indexed therein and

- access to objects thus stored in response to requests for their identification data, said stored and accessed objects and said attributes associated with said objects passing through a cache to which said object storage system and said indexing system are connected.

With the method according to the invention, there is a scalable technology making it possible to add different attributes without having to restructure the database, which is extensible, secure, fast and inexpensive. The management method according to the invention is based on a cache and advanced index structure.

It should be noted that the processed objects can also be stored in an object-oriented database.

The method according to the invention is particularly suitable for any database in which the elements are interlaced and scalable and requiring a multi-criteria search with access speed constraints. We can notably cite information systems in fields as diverse as human resources, the military field, the medical field both for administration and for research, social security, insurance, industry, education and more generally any field involving large groups of evolutionary elements.

The method according to the invention constitutes the basis of a new interleaving technology allowing access to large amounts of data linked in a complex manner, at reasonable speeds for use on the Internet, while allowing an evolution of the system with relatively low costs.

This method deals with objects which can have hundreds of attributes of all kinds including object attributes, namely self-referent or parent-referent attributes. These relationships reflect relationships corresponding to the real world. For example, a man is known to his father as 'son', to his employer as 'employee', or to his butcher as 'customer'. The interleaving technology associated with the method according to the invention makes it possible to recognize these distinctions and to act accordingly on the basis of intelligent rules. This complex data remains accessible at speeds suitable for slow networks, such as the low-speed Internet. While in a standard database system, simple data can be searched in large quantities while complex data can only be searched in small quantities, the method according to the invention makes it possible to combine these two search modes.

The object types and the attribute types remain easily extensible with the method according to the invention. In addition, while conventional database systems must be shut down with each new version or update, the engine implementing interleaving technology can be accessed via an administrative interface and updated without any programming knowledge. and without requiring a system shutdown.

The proposed interleaving technology is modular, which allows the system to be distributed over several machines to share the workload, with the additional advantage that if one of the machines fails, the other machines take over without interruption of the system.

This interlacing technology differs from previous technologies by a Kernel designed to interact with and take advantage of various external technologies, be scaled and manipulated without programming knowledge, maintain data integrity with a high level of security , and carry out all the processing in time fast enough for providing information on slow networks.

The Kernel uses rules to manipulate objects and their relationships. An engine using conventional artificial intelligence technology, for example a PROLOG engine, manages these rules, making the system easily upgradeable and robust.

Furthermore, the interleaving technology which is the subject of the present invention also has the characteristic of cooperating with standard surrounding technologies rather than proprietary ones, such as the following technologies: - XML (allowing to operate any object-oriented database system),

- Linux (allowing to run any UNIX system, even Microsoft NT®, - Post Grès (allowing to run any relational database).

The use of these known technologies allows the Kernel to interact with standard technologies which are constantly updated, fully supported and which can be managed by reasonably trained programmers.

A key point in the structure of the system's data layer is the redundancy of the stored data. The data layer has an object part and an index part: Object - the element object of the data layer stores the objects and attributes in a file structure in which each file represents a particular object or attribute. This simple file structure allows rapid delivery of data to a requesting machine. This element can be extended indefinitely with the usual hardware limitations of computer storage technology. Index - the index element stores an indexed version of object-attribute information in a simple database. This layer allows, for quick searches, the supply of identifiers necessary to the cache for the deployment of the required data to the cache / kernel.

The method according to the invention can also advantageously comprise a dynamic construction and management of directories and / or dictionaries specific to main requests, a reconstruction of a corrupted index from data stored in the object-oriented database, and an adaptation of the size and structure of the index.

It can also be provided, when a block of addresses is associated with an object, that the method according to the invention comprises a serialization of the addresses in the serialization of the object itself. According to another aspect of the invention, a system is proposed for managing information in the form of objects organized into object classes, each object comprising identification data and a list of attributes, implementing the management method according to the invention, this management system comprising a kernel and cache means.

The management system according to the invention is characterized in that it further comprises:

- means for storing objects, including for example a serialization of object files such as files in XML format, - means for indexing attributes associated with said stored objects, within an indexing system including a database relational,

- means for accessing, within said indexing system, object identification data in response to requests on attributes which are indexed therein and - means for accessing objects thus stored in response to requests on their identification data, said object storage means and said indexing means being both connected to the cache means.

The means of access to stored objects can advantageously comprise a file system.

According to yet another aspect of the invention, a method is proposed for constructing and managing dynamically, from a server connected to a communication network, pages including documents, implemented in an information management system according to the invention. invention comprising a Kernel, cache means, indexing means and object storage means.

This dynamic construction and management method is characterized in that it comprises, in response to a request for access to a site managed by said server, received by the Kernel from a terminal of a user connected to the communication network:

- a request issued by said Kernel to cache means, an object of identification of the owner of said site, - a request by said ^'cache means for extracting said identification unit, means for storing objects,

a reading from said object database of a file corresponding to said identification object and a transmission of said file to the cache means,

- a return by said cache means to the Kernel of the file corresponding to said identification object,

- processing by the Kernel of the information contained in said file, in particular to recover the identifier ID of the site sought, - a request issued by said Kernel to the cache means for accessing the site sought from said identifier,

a request issued by said cache means to the object storage means to obtain the file corresponding to the site sought,

- a reading and a return by the means of storage of objects, of the file corresponding to the searched site, to the cache means which directly send the said file to the Kernel,

- a return by said Kernel to the terminal of said user via said communication network, of all the information contained in the object corresponding to the site sought, and - a display on the terminal of all or part of said information.

According to a particular characteristic of the invention, the method of construction and dynamic management of pages according to the invention further comprises, in response to a request to add an object including a file and attributes, sent from a terminal d '' a user connected to the communication network:

- a request sent by said terminal to the Kernel, to create a new object, and a supply of all the elements constituting said object,

a request sent by said kernel to said cache means, for creating the new object, a recording by the cache means in the object storage means, of the file of said new object, and

- an indexing of the attributes of said new object by the indexing system. The method of construction and dynamic management of pages according to the invention can also comprise an emission by the cache means of a confirmation of the creation of the new object intended for the Kernel, as well as a transmission by the Kernel to the terminal. of the creation confirmation.

Furthermore, the cache means can advantageously be provided for saving the file corresponding to the new object created as soon as possible, in the data storage means such as an object-oriented database. In a preferred form of implementation according to the invention, the method of dynamic construction and management of pages according to the invention further comprises, in response to a search for an object corresponding to a set of attributes, carried out from a terminal connected to the communication network, - reception by the Kernel of search parameters transmitted from said terminal and transmission by said Kernel of a request to the indexing system,

- a supply by said indexing system of all or part of a list of identifiers corresponding to this search, and - a retransmission by the Kernel of said list to the terminal to allow display of the results of the search.

According to yet another aspect of the invention, a system is proposed for dynamically building pages from a server connected to a communication network, comprising means for connection with a communication network, a kernel, cache means, indexing means and object storage means, implementing the method according to the invention.

One can for example provide a system implemented for a site for contacting several groups of people, in which the objects treated belong to several classes of objects comprising a first class of “Media” objects relating to documents of base, a second class of "Vcards" objects modeling physical persons or moral, and a third class of "Container" objects corresponding to collections of objects.

Container objects can in particular include a “Book” object in the form of a list of attributes defining its content, and sites in the form of Book objects with specific attributes.

The system for constructing and dynamically managing pages according to the invention can advantageously be arranged to deal with relationships between objects belonging to one or other of the first, second or third classes of objects. Furthermore, it can advantageously be provided that the system further comprises means for associating with each object of one or other of the three classes of objects a cost of possession and connection. For example, a Media object can be associated with a zero cost of ownership, while a Container object can be associated with a cost of ownership equal to the sum of the costs of linking the objects it contains.

Other advantages and characteristics of the invention will appear on examining the detailed description of a mode of implementation which is in no way limitative, and the appended drawings in which:

- Figure 1 illustrates a practical example of implementation of a management system according to the invention;

- Figure 2 illustrates a site delegation mechanism in a management system according to the invention;

- Figure 3 illustrates an example of relationships between VCard and Media;

FIGS. 4 and 5 respectively illustrate examples of relationships of the Book / Media type and of the Book / Book type;

- Figure 6 illustrates a practical case of relationships within a management system according to the invention;

- Figure 7 shows an example of a class diagram according to the UML model in a management system according to the invention; - Figures 8, 9 and 10 illustrate, in the context of the management method according to the invention, relationships between objects, and between Media and Vcards; - Figure 11 illustrates, in the context of the management method according to the invention, the presence of a cache within the Kernel ,;

- Figure 12 illustrates possibilities of reconstruction within a management system according to the invention; - Figure 13 illustrates the hardware structure of a management system according to the invention;

- Figure 14 illustrates possibilities of extension ("scalability") of the management system according to the invention;

- Figure 15 is a diagram describing the operation of the dynamic construction and management method according to the invention, in the form of a timing diagram;

FIG. 16 is a first practical example of an operating chronogram of the construction and dynamic management method according to the invention, in the case of the display of a site; FIG. 17 is a second practical example of an operating chronogram of the dynamic construction and management method according to the invention, in the case of adding a Media; and

FIG. 18 is a third practical example of a chronogram of operation of the dynamic construction and management method according to the invention, in the case of a search.

The technical structure of an engine implementing the interleaving technology which is the subject of the method according to the invention comprises a set of concentric layers:

- an API layer allowing the system to communicate with other technologies, to affect data as well as to change the functionality of the system,

- a Kernel layer which secures, defines, regulates data and access to data, and defines the rules and manages the other layers,

- a cache layer which allows an optimization of the system for use on the Internet, by frequently storing the information used and providing it to the Kernel without wasting time by sending requests to the data layer, - a data layer which stores the object and attribute information to be processed by the Kernel and to be supplied to the API to be viewed.

We will now describe a practical embodiment of a system for constructing and dynamically managing pages according to the invention, with particular reference to FIG. 1. Such a system S according to comprises:

- a first central unit A providing the “Firewall” function to protect the site against any attack or external intrusion,

- a second central unit B ensuring the function of Web server and provided for calling the kernel and dynamically generating HTML pages,

- a third central unit C serving as a kernel command processing server,

- a fourth central unit D ensuring the cache function and allowing rapid access to the most used elements of the storage solution, ensuring the dissemination of information between the Data part and the Index part in order to ensure consistency of the information, and obtaining a unique reference to an object shared by several kernel instances, and

- A fifth central unit E ensuring the storage function of all the information and indexing of all the objects.

This central storage unit E hosts a Data application and an Index application and comprises a permanent storage unit such as a hard disk of sufficient capacity to store the data corresponding for example to the media and the attributes of the objects. A dynamic page construction and management system according to the invention thus comprises an architecture in three distinct main layers:

- storage: a mass memory system with interrogation functions to store and retrieve all the data entered by users,

- Kernel: the kernel provides all the basic services necessary for the dynamics of objects, including definitions and elementary rules describing the data of the model as well as the functions which allow to animate this set,

- Application: any user program of model objects.

We will now describe different families of objects used in a particular example of a dynamic page construction and management system according to the invention implemented in the field of fashion, and of the characteristic relationships between these objects, with reference to Figures 2 to 5.

For this particular domain, three families of objects are considered:

- M media, which are basic documents of the system and correspond to text, image, sound, video or other files,

- V entities, hereinafter referred to as VCard, modeling natural or legal persons,

- collections of objects, hereinafter referred to as the Container, consisting for example of lists of media or Containers, VCards, or any other object within the dynamic system. An M medium is accompanied by a series of attributes used to describe it:

- Title: summary text description of the media, - Category:

- Accreditations:

- Technique: subset of attributes describing the techniques used to create the original document,

- Environment: circumstances in which the document was drawn up,

- Atmosphere: attributes describing the document according to subjective and emotional terms,

- Legal: legal notices, rights, copyright,

- Dates: creation, validity, end of rights, - Notes: comments, free text One of the objects derived from the Container is the “Book” or “Book” B object, suitable for modeling bookmarks, address books, sites, or any aggregation entity.

Another object derived from the Container is the “site” object S, which consists of a Book with specific attributes:

- description,

- contact,

- a list of attributes from the VCard to display,

- the Container itself, which can contain media, VCards as well as Books.

We can provide for a site delegation, as illustrated in Figure 2. Thus, when A creates a site, it specifies for whom this site is intended. It is thus possible to create a site for another person B. In a second phase, B attaches to its VCard the site created by A and assigns it notions of visibility by country or by profession. A remains at all times the owner of the site he created for B.

The Media / VCard relationships expose, with reference to Figure 3, each stakeholder concerning the production of a media: the photographs

1 and 2 have a series of accreditation attributes to express the role of each in their development. From these properties, it is for example possible to deduce that:

Took photograph 1 and posed for photograph 2 B posed for photographs 1 and 2 C took photograph 2. A Book or Book object is a simple aggregation object derived from a

Container and its attributes define its content which can be any type of object, ie media or other Books or Books, as illustrated in Figures 4 and 5 corresponding respectively to the Book / Media and Book / relationships. Book. In the information management system according to the invention, certain relationships between objects require the use of specific objects, as illustrated in FIG. 8 to 11. In the first case, the relationship by its use and / or its semantics has a well defined meaning. We can move this relation into a property of one of the objects. The choice of the object depends on the cardinalities and the use which is made of this relation, that is to say of the object of origin in the majority of navigations. We preferably choose the object that allows to create this relationship with a unique property, especially if it corresponds to the usual navigation. As an example, we can consider the ownership relationship. For this relation, each object is endowed with an “owner” property whose value is the identifier of the VCard of the owner of the object, as illustrated in figure 9.

In a second case, the relation concerned has no preferential sense of use and it cannot be reduced to a property or a single attribute in one of the objects. This relation can only be materialized by a proper object, especially if this relation itself has properties and attributes which specify its nature. As an example, we can cite the participations which describe the links between a person (VCard) and a media to which they have contributed to the production. Participation defines the role of the person in the development of the media. This notion can only be modeled correctly by the use of a specific "participation" object which has three essential properties. A first property is the reference of the participant's VCard. A second property is the reference of the medium, and a third property is the role of this person during the production of the medium, with reference to FIG. 10. This can possibly be supplemented by a collection of attributes which specifies the role. A cache is put in place to access an object without worrying about its presence in memory or not, speeding up access since the objects remain in the cache until explicit removal or automatic removal due to lack of capacity in the cache, scheduling correctly entries in the database to maintain data integrity at all times.

The cache can only work if it is the exclusive source of data for the rest of the kernel, which allows it to offer a new function: database interface, the rest of the kernel no longer has to turn to the database and become completely independent of it, with reference to Figure 11. Any changes or evolutions of the database are all the easier.

The fashion construction and dynamic page management system according to the invention can be used by four main families of users:

- the surfer, anonymous user of the system, who is integrated into the model by using a default VCard card which is common to all surfers and which has a specific attribute to report his quality as a surfer and therefore prohibit access to objects reserved for professionals and any manipulation of system objects,

- the guest, who is a surfer with extensive rights and who has a VCard with a specific attribute to ensure the processing of their particularities,

- the member, who can create objects and access all the services offered by the site, except for sites subject to access restrictions imposed by the members themselves via feasibility attributes, - the administrator , who can use the system like any user and who is responsible for maintaining the system and can possibly act as mediator in the event of a conflict between members. In a dynamic page construction and management system according to the invention, numerous situations of communication and information exchange between users of the site using this system can be envisaged, as illustrated in FIG. 6 corresponding to a case practice involving for example a photographer "Marc", a Model "Claudia" and an Agency. The management system contains, for this practical case, eight photographs in the form of eight M1-M8 Media. The photographer, holder of a VCard V (Ma), took the photographs M1, M2, M3, M6, M7 and M8 and personally loaded the photographs M1 and M6 into the system. He created a Book or Book B (Ma) of photographs of the Model containing the photographs M1 and M2. In addition, the photographer has a site S (Ma) in which is published Book B (Ma) and photography M3.

The Agency, holder of a VCard V (E), loaded into the system according to the invention the photographs M3, M4, M5 and M7, and created a General Information Book Ea containing the photographs M4 and M5. She also created a Book Eb of the model containing the photographs M6 and M7. This agency also has a site S (E) in which the Books Ea, Eb and the photograph M3 are published, and has created a specific site C1 for the model, which contains the Book Eb and the photograph M8.

The Model posed for the photographs M1, M2, M6, M7 and M8, and loaded into the system according to the invention the photograph M8. In addition, the Model has a C0 site intended for the general public GP, created by itself and containing Book B (Ma). In addition, the Model defined that the C1 site created by the Agency was intended for French professionals. We could also have defined a restriction by trade. The action of loading a photograph into the management system according to the invention has the effect of creating a Media object whose owner is the user who carried out this operation.

It should be noted that, in order not to overload the diagram in FIG. 6, the accreditation links between the different protagonists are not represented. As an example, photography M7 defines by its attributes three of these links: to the photographer, to the Agency and to the Model.

Each object of the dynamic system can have a cost of ownership, a cost of connection and generate gains. An invoice is drawn up periodically on the basis of events (object creations, object connections) which influence the cost. It is possible to know precisely, for each object, the history of these costs. Thus, the objects can be established taking into account the actual duration of the use of these objects. By way of nonlimiting example, one can consider the following costs of ownership and connection:

- media has a zero cost of ownership,

- a user can insert as many media as he wishes in the dynamic system according to the invention without this affecting his bill,

- the cost of owning a Container is the sum of the costs of linking the objects it contains, the cost of linking being arbitrarily fixed at a fixed value according to its type (book, site, ...).

As illustrated by the class diagram in UML model represented in FIG. 7, all the objects implemented in the management system according to the invention have certain common characteristics, including those of having an identifier and a proprietary attribute whose value is the VCard of the user who introduced the object into the management system according to the invention. In addition, it is possible to enrich the library of attributes at any time by deriving the existing attributes to specialize or specify them, or to create new ones by directly deriving the Attribute object.

We will now describe, with reference to structures 12 to 14, an example of the internal structure of the information management system according to the invention. This management system comprises a kernel containing an API interface implementing the management method according to the invention and driving on the one hand a database for storing media and on the other hand a database for indexing the attributes associated with the stored objects. Each of these separate databases is associated with data and index file systems, respectively. It should be noted that the independence of indexing from storage and the high redundancy induced by this particular mode of management makes it possible to reconstruct a corrupted index from the data already recorded. The components of the dynamic construction and management system according to the invention include, with reference to FIG. 13:

- a storage and indexing layer D, - a layer of cache C,

- a core layer or Kernel K,

- an API layer of the kernel, which can have several forms such as a CORBA, RMI object, a command interpreter accessible by a socket,

- a web application layer W, including a generator of HTML pages from the kernel services accessible via the API of said kernel,

- an HTTP transport protocol layer, - a web browser layer for the user.

Furthermore, a dynamic construction and management system according to the invention preferably comprises anti-intrusion security elements such as a “firewall” or firewall, to prohibit malicious external connections. In a practical embodiment, a dynamic construction and management system according to the invention can be produced from five separate machines, as illustrated in FIG. 1 above.

The independence of the different layers of the construction and dynamic management system according to the invention makes it possible to locally adjust the necessary resources, until it becomes a fully distributed system where each layer has several machines as illustrated in FIG. 14.

We will now describe, with reference to FIGS. 15 to 18, an example of the operation of a dynamic page construction and management system according to the invention, implemented in the field of fashion. In the following, we will use the letters W, K, C, D and I to designate respectively the Web, the Kernel, the Cache, the Data and the Index, and we will describe examples of operation in the form of chronograms such as the one shown in figure 15.

In a first example corresponding to the case of a user wishing to access a personal GP site of a member, in this case the Claudia Model, illustrated in Figure 16. In (16.0), W asks the Kernel the elements allowing to constitute the requested site. The Kernel then requests (16.1) from Cache the VCard object of the owner of the requested site, in this case Claudia. If we assume that the Cache does not have this VCard at this time, the Cache will therefore request it (16.2) from database D which then reads (16.3) the corresponding file, without any search, and returns it to the Cache. The Cache, having no processing function, sends the VCard directly to the Kernel (16.4), which then analyzes the information contained in the VCard (16.5). In particular, the Kernel searches there for the site identifier ID corresponding to the GP site sought which is assumed to be, for example, S2589. The Kernel then requests from Cache the site identified by S2589 (16.6).

We also assume that the Cache does not have this object at the moment: the Cache will therefore request it from the database D (16.7) which then reads the file and returns it to the Cache (16.8). The latter directly refers the Site to the Kernel (16.9). The latter then sends all the information on the Web, which includes the VCard and the Site, and is then displayed on the screen of the user's computer or terminal. The list of Media IDs and the Description of these Media, which are contained in the Site object, can also be displayed without further request.

In a second example corresponding to the case of a user wishing to add media, illustrated in Figure 17, the Web transmits (17.0) a request to the Kernel to create a new media and provides it with all the necessary elements (attributes and file ). The Kernel then requests from Cache the creation of the new object (17.1). The Cache then confirms the creation of this new object (17.2), then it is up to the Kernel to confirm to the Web the creation of said new object (17.3). The Cache then saves the file corresponding to the media as well as the serialization of this object (attributes) in a file (17.4) as soon as possible. The attributes of the new object are then indexed by the indexing system in order to be able to perform searches (17.5). In a third example corresponding to the case of a user searching for objects corresponding to a set of attributes, illustrated in FIG. 18, the Web transmits the parameters of the search to the Kernel (18.0), which then makes a request to the system indexing I (18.1). The latter returns the list - or part of this list - of ID identifiers corresponding to the search (18.2). The Kernel then retransmits this list on the Web to allow the display of the list of results.

Of course, the invention is not limited to the examples which have just been described and numerous modifications can be made to these examples without departing from the scope of the invention. In particular, other families of objects than those which have just been described can be envisaged depending on the fields of application concerned. Furthermore, for the serialization of object files, it will be possible to use, in the remainder of the XML technology, other future technologies for describing and modeling objects.

Claims

1. Method for managing information modeled in the form of objects organized into object classes, each object comprising identification data and a list of attributes, characterized in that it comprises:

- access to objects thus stored in response to requests for their identification data, said stored objects and accesses and said attributes associated with said objects passing through a cache to which said object storage system and said system are connected. indexing.

2. Method according to claim 1, characterized in that it further comprises a dynamic construction and management of directories and / or dictionaries specific to main requests.

3. Method according to claim 1, characterized in that the objects are stored in an object-oriented database.

4. Method according to claim 3, characterized in that it further comprises a reconstruction of a corrupted index from data stored in the object-oriented database.

5. Method according to one of the preceding claims, characterized in that it further comprises an adaptation of the dimension and the structure of the index.

6. Method according to one of the preceding claims, wherein a block of addresses is associated with an object, characterized in that it further comprises a serialization of said addresses in the serialization of the object itself.

7. Method according to one of the preceding claims, characterized in that it further comprises a storage of objects in a container which is loaded into memory only on explicit request.

8. System for managing information modeled in the form of objects organized into object classes, each object comprising identification data and a list of attributes, implementing the management method according to any one of claims previous, this management system comprising a kernel and cache means, characterized in that it further comprises: - means for storing objects in a serialization of object files such as files in XML format,

means for indexing attributes associated with said stored objects, within an indexing system including a relational database,

means for accessing, within said indexing system, object identification data in response to requests for attributes which are indexed therein and

- Means for accessing objects thus stored in response to requests for their identification data, said object storage means and said indexing means being both connected to the cache means.

9. Management system according to claim 8, characterized in that the object storage means comprise an object-oriented database.

10. Management system according to one of claims 8 or 9, characterized in that the means of access to objects comprise a file system.

11. Method for constructing and dynamically managing, from a server connected to a communication network, pages including documents loaded by users, implemented in an information management system according to one of claims 8 to 10 comprising a Kernel, cache means, indexing means and object storage means, characterized in that it comprises, in response to a request for access to a site managed by said server, received by the Kernel from a terminal of a user connected to the communication network: - a request sent by said Kernel to the cache means, of a characteristic object of identification of the owner of said site,

a request issued by said cache means for extracting said characteristic object from object storage means,

a reading within said object storage means of a file corresponding to said characteristic object and a transmission of said file to the cache means,

- a return by said cache means to the Kernel of the file corresponding to said characteristic object,

- processing by the Kernel of the information contained in said file, in particular to recover the identifier ID of the site sought,

- a request issued by said kernel to the cache means for accessing the site sought from said identifier,

- a request sent by said cache means to said object storage means, to obtain the file corresponding to the site sought, - a reading and a return, by said object storage means, of the file corresponding to the site sought, cache means which send the said file directly to the Kernel,

a return by said Kernel to the terminal of said user via said communication network of all the information contained in the object corresponding to the site sought, and

- a display on the terminal of all or part of said information.

12. Method of construction and dynamic management of pages according to claim 11, characterized in that it further comprises, in response to a request to add an object including a file and attributes, sent from a terminal of a user connected to the communication network: - a request sent by said terminal to the Kernel, to create a new object, and a supply of all the elements constituting said object,

- a request issued by said kernel to said cache means, for creating the new object,

a recording by the cache means in the data storage means of the file of said new object, and

- an indexing of the attributes of said new object by the indexing system.

13. A method of construction and dynamic management of pages according to claim 12, characterized in that it further comprises a transmission by the cache means of a confirmation of the creation of the new object to the Kernel.

14. A method of construction and dynamic management of pages according to claim 13, characterized in that it further comprises a transmission by the Kernel to the terminal _, of the creation confirmation.

15. Method of construction and dynamic management of pages according to one of claims 11 to 14, characterized in that the cache means are provided for saving as soon as possible the file corresponding to the new object created, in the storage means of objects.

16. Method for constructing and dynamic management of pages according to one of claims 11 to 15, characterized in that it further comprises, in response to a search for an object corresponding to a set of attributes, carried out from a terminal connected to the communication network, a reception by the Kernel of search parameters transmitted from said terminal and a transmission by said Kernel of a request to the indexing system,

a supply by said indexing system of all or part of a list of identifiers corresponding to this search, and

- a retransmission by the Kernel of said list to the terminal to allow display of the search results.

17. System for dynamically building pages from a server connected to a communication network, comprising means of connection with a communication network, a kernel, cache means, indexing means and object storage means , implementing the method according to any one of claims 9 to 14.

18. dynamic page construction and management system according to claim 17, implemented for a site for bringing together several groups of people, characterized in that the objects treated belong to several classes of objects comprising a first class d 'Media' objects relating to basic documents, a second class of 'Vcards' objects modeling natural or legal persons, and a third class of 'Containers' objects corresponding to collections of objects.

19. The system of claim 18, characterized in that the container objects include a "Book" object in the form of a list of attributes defining its content.

20. The system of claim 19, characterized in that it further comprises sites in the form of Book objects having specific attributes.

21. System according to one of claims 17 to 20, characterized in that it comprises in that it is arranged to process relationships between objects belonging to one or the other of the first, second or third classes objects.

22. System according to one of claims 17 to 21, characterized in that it further comprises means for associating with each object of one or other of the three classes of objects a cost of possession and connection .

23. System according to claim 22, characterized in that a media object is associated with a zero cost of ownership.

24. System according to one of claims 17 to 23, characterized in that a container object is associated with a cost of ownership equal to the sum of the costs of linking the objects it contains.

25. Application of a system according to one of claims 17 to 24 for managing and providing press books or "Pressbooks" on the Web to the various players in the fashion field.