KR19980079721A - Method and apparatus for optimizing use of circuit resources for information request in information network - Google Patents

Abstract

A new type of service has been disclosed for handling user requests in an information network, such as an intranet or the world wide web, within a company, which includes limited line resources that cannot accurately locate the information to be retrieved in advance. It can be deployed in an information network. This service is arranged in the circuit network to facilitate the backside processing or batch processing of the request. According to a first optimization step, the service is responsible for executing the search task independently from the user in response to receiving the request. By executing the process behind the scenes, there is the possibility of controlling the start time or timing of user task processing in such a way that the overall line cost is minimized. According to a second optimization step, it has been proposed to separate the processing of the work by the service itself through processing the request via the information network.

Description

Method and apparatus for optimizing use of circuit resources for information request in information network

The present invention relates to a method and apparatus for processing a request in a client / server-based information network, wherein the server is an intranet or Internet within a company. Performs a task on behalf of the client, and the request is generally sent from the client to the server via an overly dense line connection. The present invention can be preferably used for extracting information in an information network, especially in a network that cannot accurately specify the location of the information in advance.

As the Internet's most popular service at present, the World Wide Web (WWW) has recently experienced significant advances in all kinds of information fields. Internet users surf the network and collect the information they need very easily. However, since the required line connection is not faster than the slowest link in the chain of connections between the user and each source of information he seeks, such surfing is generally considered to be very time-consuming even in times of high-speed transmission facilities such as ISDN. . One of the reasons for this phenomenon is that in the case of the so-called TCP / IP (Transmission Control Protocol / Internet Protocol) communication protocol, the actual information route is not generally predictable, and thus the overall circuit. The fact is that the throughput of data is reduced by the inferior portion of the connection.

For this reason, the end user of the Internet, on the one hand, bears a considerable cost to maintain the necessary dial-up connection to the network provider. On the other hand, in surfing, it takes quite a long time for the information to be found ultimately appearing on the user's screen.

In addition, the Internet is increasingly used as a communication medium in the field of business. According to a report referring to Internet business as an example (WWW User Report, April / May 1996) (Verlag Neue Mediengesellschaft. Publisher W. Deyer H. Summa), business customers now make up half of Internet users. In this field, the typical high response time required to retrieve the information is clearly unacceptable. At this point, as noted above, the net transmission rate is basically determined by a bottleneck in the entire transmission chain, so that faster line connections between the Internet provider and the Internet generally allow for higher data throughput. Emphasize that you do not let them. In special cases where the bottleneck has its origin in the network provider itself, the transfer rate can be significantly increased by generally increasing the circuit connection, but this obviously incurs an additional cost burden on the network provider, and ultimately the end user. Incurs additional costs.

The network provider's primary goal is to optimize the use of Internet access anyway, so for the network provider, techniques for minimizing or avoiding data transmission peaks and as much as possible daytime network use are as even as possible. Significant attention is paid to techniques for decentralization. Thus, data traffic that proceeds as uniformly as possible over time is another concern of the network provider with network access constantly affected at optimal rates.

In addition, users of the Internet generally have little know-how about the data traffic running behind the network. They only face relatively low throughput rates without knowing more precisely about their background or cause. Thus, in addition to the services made available to a portion of the network provider, there is a need for additional services that will result in an increase in the data throughput for both the network provider and the end user. Regarding this kind of service, it is clearly assumed that there is a special knowledge of network topology and data transmission traffic, and its implementation in the case of data traffic.

In the user-specified request area of the information network, various approaches already exist to manage the flow of huge amounts of information. These are based on the knowledge that surfing on the Internet is very time intensive and that the information received is generally not necessary for each user.

One of the first approaches was achieved by what is called an intelligent agent, which thoroughly examines the available information 'on-line' for items of information specifically needed by the user. It is. Suitable agent programs are described, for example, in Neural Intelligent Agents, a paper by D. Jeaps, which has already been published in the Online CD ROM Review (Vol. 20, No. 5, pp. 260-262). Here, we describe an autonomous program based on a neural network. In this system, the user communicates information with the agent, for example, about a desired subject area before starting a new search.

A similar approach for the current type is SW Loke's paper CiFi: An Intelligent Agent for Citation Finding on the World Wide Web (Springer Verlag Berlin, 1996, conference summary for PRICAI '96: Topics in Artificial Intelligence 4th Pacific Rim International Conference on Artificial Intelligence. Here, a rule-based agent for the world wide web has been proposed, which is located on the user side when looking for information. These agents use the spectrum of search strategies to search and find web pages completely autonomously. Here, the heuristic search method is the forefront, and the search begins first with the LYCOS search engine, which is well known on the world wide web.

Further agent-based methods in the field of database access for different types of databases are described in European patent applications EP 0747841, EP 0747843 and EP 0747844. The intelligent agent described there plays an arbitration role between the client and the server, and the server handles the tasks related to the request completely at the request of the client. To be sure, special attention is given to minimizing the daily tasks required on the user's side and to precluding or at least greatly reducing those interventions by the user. The proposed solution there is to provide an executive program agent. For example, a user's request processed by a web server is first delivered to the management program agent, which in turn is responsible for handling the request in the network by a sub-agent. After performing the appropriate task, the management program agent again returns the entire result to the client at the address specified in the original request for the correct display format. This procedure ensures that the user presents only one task, where the actual task is performed by the agent behind it.

Although the above solution introduces a number of features that facilitate the user searching for information in local and / or distributed information networks, the solution still does not take into account the time factor of the request entered by the user. Without the time dependent cost, such as all charges incurred for the transmission line network.

These solutions also have the disadvantage that the different transmission capabilities of the line connection between the end user and the information network are not taken into account. Between the end user and the network provider there are mainly line connections with extremely different transmission bandwidths, such as for example ISDN or modem connections to a regular switching network. Thus, there is relatively little or no cost to operate a circuit connection between the network provider and a server set up in an information network, while the connection setup between the end user and the network provider is usually relatively Become cost-intensive.

Accordingly, the present invention is based on the object of providing a method and apparatus of the above-described type, which avoid the above-mentioned disadvantages, and in particular allow for the above-mentioned time and cost problems. In addition, the basis according to the invention should, on the one hand, be made so as to optimize the access network to the information network in terms of load distribution. On the other hand, according to the present invention, it is possible to continuously utilize the Internet combination between one network provider, intranet or end user and the other side of the Internet as possible. In addition, optimal resource utilization can be seen between the process executed when the user's request is processed and the communication means required to send the job to the Internet.

This object is achieved by the features of the independent claims of the method and apparatus. Another preferred configuration is the inventive idea of the individual dependent claims.

1 shows a prior art information network which is the background of the present invention;

2 is a conceptual diagram of a service by a first optimization stage of the present invention;

3 is a conceptual diagram illustrating a second optimization stage of the present invention in comparison with FIG.

4 is a diagram corresponding to FIG. 1 in which a service according to the invention is additionally inserted between a network and a network provider;

5 is a view showing another embodiment of the present invention in which a service according to the present invention is disposed in a network provider itself, and corresponding to FIGS. 1 and 4;

FIG. 6 shows that the retrieval of a document is performed behind the scenes according to the user's request according to the present invention.

7A and 7B are flow charts of additional processing operations based on a reference between existing documents and a flow chart of functions performed when a job is processed by the agent of the present invention.

8 is a conceptual block diagram of a preferred embodiment of a second optimization stage according to the present invention.

9 is a conceptual diagram of the temporal evolution of the acquisition and analysis process of a second optimization stage according to the present invention.

10 is a diagram of an embodiment of an object model (described by UML (Unified Modeling Language)) of the agent in which a basic class for job processing is represented.

11 illustrates a preferred form of performance of an object oriented task scheduler with a second optimization stage of the present invention.

12 is a detailed summary of interactive job processing in accordance with the present invention.

* Explanation of symbols for the main parts of the drawings

1, 2: line connection

In the environment of the first optimization stage according to the present invention, a commercial service disposed in a circuit network is provided, and the user request in the form of batch processing in the network is subjected to back processing. (background processing). When the request is forwarded to the service, it is dedicated to the execution of the task-independent (not connected) from the user. Since the processing is performed behind the scenes, the possibility arises of controlling the start time or timing of the user's work processing, in particular in such a way that the line cost is generally minimized.

By means of the second optimization stage, the present invention essentially does not convey more demands to the Internet than can be handled by limited bandwidth, so that it can take advantage of continuous Internet coupling as much as possible without causing an overload situation. Shows that Information that satisfies the conditions specified on the user side (e.g., document) is therefore not immediately required, but a list with a destination address, for example, document address (URL), is added. If the surplus bandwidth for processing is fully available, this list or pool of addresses is used to create or set up new acquisitions and analysis.

In more detail, the processing of user requests, such as, for example, the analysis of already available information (search for additionally included references) and the need for additional information from the Internet based on the results of the already available analysis, are independent of each other ( Separately). For this purpose, interim results, such as found references (URLs), are temporarily stored in a pool (buffer). By these means, for example, even if there is no surplus transmission capacity (bandwidth), it is possible to make the processing of the user request especially possible to be carried out continuously. Thus, all available surplus transfer capacity can be optimally used in that the work address already temporarily stored in the buffer can be used immediately for future work processing. By not combining as described above, the continued use of the Internet connection is facilitated.

The service according to the invention is especially for the needs of network providers and their customers (end users) and for intranet operators. In addition, with the help of these services, the cost on both the supplier side and the user side is significantly reduced. This is made possible, in particular, by the processing of user requests being performed behind the scenes (offline) instead of the on-line retrieval so far required. Processing work behind the scenes makes it possible to use optimal time schedules, in particular for the transmission of information, for example, both inside and outside of the information network between the network provider and the end user. Significant cost savings are achieved by the optimum throughput ratio at the connection points of the network achieved as described above. The information required by the customer is searched and collected in a time when the network use of the information network is less by the service according to the present invention, and is only transferred to the end user as the inquiry is completed.

The collected information can only be transferred in accordance with a further embodiment of the present invention upon application by the user, while the information is temporarily stored by the proposed service itself in the meantime.

If the security of the communication is sufficiently maintained, the information may be sent directly to the end user, for example in the form of an FTP upload.

In addition to one operation, continuous operations such as extracting the current stock market price from a server during the week, for example, may also be allowed, and may be allowed to the client (end user) as an information packet. Can be made available.

Another advantage of the backside processing is that it relieves end users of the burden of user input that still needs to be handled interactively, for example in the case of current web browsers. In the case of the present invention, this method of conversation is replaced by a task, which includes the search term inter alia, search depth, if possible, for example, to be specified first. Search engines, time limits, quantity limits, other protocols for the protocol to be used (e.g., NNTP for news, FTP for file archives), embedded images to be loaded, rules for sound, etc. do.

Regarding the apparatus by the first optimization stage of the present invention, an appropriate agent is placed in the circuit network. In a preferred embodiment, the agent is arranged at an interface between, for example, a server of each service provider, a connection to a circuit network with a relatively wide transmission bandwidth and a connection to a circuit network with a relatively narrow transmission bandwidth.

Another embodiment of the agent relates to a transport service based on a modem connection. In this kind of overview, it is preferable to place the agent in the line connection area between the end user and the modem service. This arrangement is particularly large circuit caused by the low rate of modem connection, since the information transmitted by the agent to the end user has already been injected in a compact form of the kind requiring as little transmission capacity as possible. The cost is minimized very effectively.

With regard to the method and apparatus of the present invention, it is possible to additionally provide a result of a search that can be supplied back to the end user as a data packet that can be examined by the agent, which includes HTML (hyper text markup). Hypertext Markup Language (Hyperlink) is also included in the document description language. Thus, the end user will be able to surf locally on the transferred information and automatically eliminate the shortcomings of large response times in interactive work with the Internet. Hyperlinks to unsupplied documents are here where they once again point to the Internet and the user can enter the Internet directly and bypass the network if necessary to read other documents not obtained by the search criteria. Modified by downloading them.

In addition, in the case of the present invention, a filter function may be provided that can significantly limit the scope of information found by the agent.

According to another embodiment, a bandwidth reservation may be provided to optimize utilization of transmission resources of the information network.

Further, according to another embodiment of the present invention, the fact that the information obtained by the agent and transmitted to the user is minimized by a filter, and its parameters can be defined when the user requests it. Thereby, the transfer capacity available both between the information network and the agent and between the agent and the end user can be better utilized.

In addition, in order to optimally use the transmission resource between the agent and the user, it may be further compressed before transmitting the information to be transferred between the agent and the user. This type of performance is particularly applicable in the case of modem connections with relatively little margin of transmission generally between the user and the network provider or agent.

The user's request can be delivered to the agent in the form of a (task), which includes information required for backside operations or batch operations. There, both the criteria for the information to be observed and the parameters for filtering the information to be transmitted can be specified by the user. In the information to be searched, a boundary thereof can be specified in the form of a search word or as a direct search address (for example, a URL).

In another embodiment, the agent may work with one or more search engines set up in the information network, for example, to organize search in particular efficiently. Information regularly used may be pre-stored here as the agent (cache operation). In particular, at the request of the user, several search engines can be instructed to work in parallel (simultaneously). Thus, the yield of the requested information can be increased significantly, and the search can be accelerated by the fact that the query can be truncated as the first information arrives at the agent. The arrival of information at the agent at different times can be caused, for example, by different processing speeds of addressed search engines.

In another embodiment, at least one new program thread may be provided as soon as excess bandwidth is available-in the second optimization stage.

As a desirable performance, a time scheduling module (scheduler) is provided as an additional process means, which checks the available transmission bandwidth at regular time intervals, and provides a new program. Appropriate access to other process means and buffer stores to create threads.

In addition, in the case of retrieval also placed in the buffer storage device, in order to ensure that the analysis and retrieval process can be timely terminated, which will continue indefinitely, especially if it does not have appropriate truncation criteria, the acquisition of the referenced document Additional information about the depth may be provided.

In addition, all customer jobs, in particular all newly incoming jobs, can be temporarily stored in the buffer storage with all the filtering conditions of the individual retrieval jobs.

In a particularly preferred embodiment, all the components required to implement the invention, in particular the functional parts relating to the core of the document request, are provided in object-oriented technology.

The invention is described in more detail below with reference to the accompanying drawings in which preferred embodiments thereof are shown.

1 illustrates a first area of an information network, here the Internet, which forms the basis of the present invention, wherein a number of end users have access to the information network through a network provider. Each situation here is basically characterized by two transmission sections, on the other hand, typically in extreme overloads of peak times and in extreme underloads of other times. There is a transmission section between the network provider and the Internet characterized by, and on the other hand there is a separate connection between the end user and the network provider whose transmission attribute is often opposite. Compared with the communication link between the network provider and the Internet, the circuit connection between the network provider and the end user is typically due to extreme underload during peak times and data throughput rates well below the available bandwidth. The feature is erased.

2, two optimization stages according to the present invention are first compared. In the first optimization stage, the service of the present invention is used to optimize the circuit connection 1, in particular by the fact that it uses as little as possible the circuit connection 1 between the network provider and the end user. This is achieved in principle by the fact that the creation of the user request and its processing are handled separately from each other. In contrast, in the second optimization stage, the line connection between the network provider and the information network (eg, the Internet) is optimized. This optimization is based on the concept that in this embodiment the functional unit set by the network provider processes the task in parallel in time and sends a request to the information network. Thus, a 'pool' of the address of the document can be generated in advance, which can be continuously serviced under the optimal line condition to the Internet. Thus, the above object is achieved in that the redundant transmission capacity is immediately available and does not require waiting for the address of a new document from the analysis process first. For example, in request processing, priorities are placed on tasks placed in the pool, such as giving documents that do not contain pictures to documents that contain pictures, or giving priority to documents that are smaller than larger documents. May be given.

Referring to FIG. 3, the form of an embodiment of the present invention is shown when the functional components required for the two optimization stages are set directly to the end user. As described in detail below with respect to the service according to the invention, the advantages caused by circuit optimization have already become useful.

The first optimization stage of the present invention is based on the concept of providing an agent in charge of the task of searching through the information network, collecting the required information from the individual end user and transferring it to the end user. The request of the user is performed in the form of background processing or batch processing by the agent in an offline state from the user. The invention makes it possible to perform retrieval and collection of information in the information network according to a schedule, wherein the execution of the request of the user is only temporarily or after creation of the request task by the user. It is only affected for a certain time. The batch processing executed on the back side is preferably performed when the information network is at least not fully used. In this way, the use of a data stream flowing in the information network is optimized on the one hand and on the other hand the traffic peak of the transmission is significantly reduced. By reducing the transmission bandwidth required for access to the information network, the operating cost is also ultimately reduced.

In the case of dial-up connection, a cost advantage has already been achieved as the present invention separates the processing of the task and the user in time and space by the agent on which it is based. After the user transfers his request to the agent, the telephone connection between the two can be interrupted. Then, it is only necessary to re-establish a new dial-up only if the collected information presented to the agent after completion of the job processing can be transferred to the user. Thus, the cost advantage arising in this regard is based on the fact that the need to maintain a dial-up connection is now removed to require the transmission of the retrieved information to the user.

The above separation also significantly improves the temporary connection performance with respect to the information system of a mobile system such as, for example, a laptop computer, compared to the prior art. Here too, after the agent has been instructed to search, the mobile system can completely disconnect from the information network. Only at a later time when the work is believed to have been processed by the agent, the connection to the information network is reestablished, and the agent sends the information to the mobile system. Thus, the mobility of such a system is significantly improved above all else.

During interaction with the Internet, the end user must thoroughly and interactively research the information network online to retrieve current information. In the first step, in order to obtain an overview of all the documents presented in the observation area of the Internet, the so-called search engine is addressed. Using hyperlinks supported in HTML documents, it is possible, above all, to extract more in-depth and detailed information from each original server with respect to the information obtained from the information provided by the search engine. This general multi-step process takes a tremendous amount of time to ultimately obtain the desired document and is therefore rarely available in the business world.

According to the service of the present invention, the search steps are performed autonomously on behalf of the end user, and at the end of the search process, the entire information is transferred to the end user in the form of one packet. Since this information packet provides references corresponding to the original hyperlink, the user is open to further search for more specific information through the limited information packet.

With regard to the network provider's access to the information network, it is also optimized and is particularly advantageous for setting up services according to the invention in the vicinity of the network provider. In the first embodiment of the invention shown in Figure 4, this service is located in the transmission line area between the Internet and the network provider, for example implementing the control function according to the function by the specific hardware described above. You can also think of other implementations. The service that collects the information desired by each user is completely autonomous and the found information is transferred directly to the user or temporarily stored so that the information becomes available as soon as the end user requests. In all cases, as the batch ends, all data becomes available to the network provider, and no kind of data need to be extracted online via the Internet, so the connection between the user and the network provider is almost anyway. It operates at the optimum transmission speed. According to the service of the present invention, it is possible to schedule the search job to be processed in more detail, thereby effectively preventing a plurality of users from simultaneously accessing the Internet online at the same time, thereby preventing the simultaneous and high-quality simultaneous access that was previously required. No need to provide

Here, the service according to the invention is located between the client and the server, not behind the server-for example, as in the European patent application cited earlier in this specification. In addition, the existence of the service (agent) should be known to either the server or server script, since it must actively inquire about it. The agent presented here is known only to the client, and to other servers it appears to be just a normal client, such as a web browser.

In comparison with the embodiment shown in FIG. 4, it would also be reasonable to provide the network provider itself with the service according to the invention if the network provider had a more preferred starting condition in order to optimize network traffic. Can be. A suitable embodiment is shown in FIG. In this embodiment, the execution program required for the service of the present invention and the necessary database or temporary storage device are arranged in the Internet provider itself, for example, the server of the network provider.

The most commonly used protocol on the Internet (or the World Wide Web) is the so-called HTTP protocol (hypertext transfer protocol). This protocol is commonly used to send HTML documents. However, this communication protocol is also suitable for transmitting all kinds of information with a so-called URL (Uniform Resource Locator). Hypertext documents include text, so-called links, and data embedded in individual documents. Thus, such a document can be considered a document network, where documents reference other documents, which can include almost any type of information such as image, sound or text information.

The data embedded in a document is first presented with each of the above documents, but for access to this data, each time a new HTTP request (task) is required to load this data. The referenced documents are sent for their portion over the network using the HTTP protocol. The HTTP web server can now inform the file type (content type) of all the information provided.

According to another embodiment of the service (agent) according to the present invention, it is proposed that a search operation completes a form defined in detail before the user (client) starts the actual search. This form may be provided, for example, by a web server connected to the agent. If the user already knows the address of the information to be searched for, the direct Internet address (URL) can be specified in the form, or if applicable, one or more search terms that are logically combined. Can be. If a search term is used, the agent will first consult at least one search engine configured on the Internet.

In order to specify a particular subject area or expense for the entirety of the information found, the upper limit of a search hit that will subsequently be notified to the user is, for example, each subject. The first ten search results for the area may be displayed in the work form. In addition, a search depth for the query may also be displayed, i.e., an upper limit may be set on how often the server should track the links of documents or sub-directories in an FTP (File Transfer Protocol) archive. In addition, the flow of information can be further limited by limiting the retrieval time or the time for downloading information from the network, and by limiting the amount of information itself.

Optionally, the agent can more precisely define the protocol and medium to be used during the query and the type of data to be retrieved. Thus, notification of image data, which requires a relatively large transmission capacity for its transmission, can often be omitted. The protocol, on the one hand, is mostly contained in hypertext documents and relates to links to other documents or websites. These links can be HTTP or FTP links, which can be recursively connected to other HTTP links. On the other hand, with respect to the protocol, it is conceivable that not all of the information found in the global information network is HTTP based. The search routine used in the technical idea of the present invention may be used in addition to this to extract the current daily news from the news server.

In contrast to HTTP documents, because the file type or content is unknown in advance, FTP documents cannot be inspected, and searches cannot be initiated within FTP archives.

One embodiment of the invention in the area of document retrieval is illustrated in detail in FIG. In the underlying search task, above all else, in this example, the search is performed in consideration of HTTP and FTP links, at search depth 2, where only documents up to the second generation (ie all two levels in the network structure) are considered. Indicated. Furthermore, only media type documents such as 'HTML' and 'PostScript' are searched.

All information obtained from the network is first temporarily stored in a file. References to other documents or embedded files included in each document are rewritten here to point to local copies within the file. For files not downloaded from the network, for example files outside the defined search depth, maintain the original link to allow the client (user) to make other queries of deeper depth based on this link. .

The flowchart shown in FIG. 7A shows the functional performance of job processing by the agent of the present invention. The search task reaching the agent is first stored in the task list. For each job currently being performed, the illustrated processing step is executed by the above embodiment. If one task has already reached the end point and can be completed, the task is first removed from the list of active tasks. The information obtained by each job is temporarily stored at the same time. In this embodiment, the stored search results are then notified by the agent to the client at the request of the client. In this embodiment, in order to optimize the communication link between the agent and the user, an assumption is made that a collection queue defines a range of tasks to be processed by the agent in advance.

Until the operation can be completed, the steps shown on the right side of the drawing are performed. First, for each partial task, it is checked whether the bandwidth required for processing the partial task is available. Consider reserves for future interactive processes. In addition, additional processing (such as the maximum processing time or maximum amount of information defined in the request profile) and the maximum number of program threads available for each agent or job may be further processed by a filter. Search) to see if the process is allowed. Depending on these conditions, the search operation may then end. A new program thread is started and the document to be retrieved is obtained. At the same time, these documents are removed from the document list.

Subsequent additional tasks, the processing of the retrieval job, are assigned to the agent. In other words, the transfer of information from the already started network to the temporary storage of the agent is not interrupted, since this may result in loss of data and may need to restart the transfer of the entire data. Therefore, no new acquisition operation is started until the required transmission capacity is available. Thus, the agent regularly checks the current situation for each available bandwidth and then requests additional documentation (independent of its own program thread) without jeopardizing the interactive work of other customers. Determine if you can.

Execution of the program thread started to obtain the document is described in detail in the flowchart of FIG. 7B. If a document does not have a reference to another document, the document is first compressed for easy notification by the agent to the client and then added to all existing archives. If there are references, especially references that have not yet been processed, each case requires a description of the referenced document. If the description of the document reveals that this document does not satisfy the existing filtering conditions according to the required profile, the reference is taken into account externally, i.e. a possible relative address is a separate external source of the Internet. (E.g., URL). However, if the document satisfies the filtering conditions, the existing hyperlink is rewritten to a reference to the document and then made available locally (as a local source). Further, reference to the document is described in the document list of the individual task in order to make the appropriate document available to the user as well.

In a further embodiment of the invention (not shown), the agent performs a dynamic reservation of bandwidth. This means that the interruption in the process of downloading information is disadvantageous for the reasons described above, and therefore, not to continuously operate available line resources at its upper limit, but to allow a new discovery process to be started, or in progress. The process is based on the recognition that it is convenient to make the circuit reservation available when data throughput is increased. This kind of reservation is also required when downloading very large linked files, because interrupts in this case are particularly negative. Available bandwidths are predictable in some frameworks because the changing usage habits of individual Internet users have only a statistically small impact in terms of large groups of all users. In addition, the usage habits of various regular Internet users only change relatively little in time.

From past and present user behavior, one can estimate the bandwidth to be used for interactive traffic immediately after. By the difference between this expected load and the maximum available bandwidth, it is possible to determine the surplus bandwidth for the backside or batch operation of the agent. However, some of this surplus bandwidth is not used by it. This part is intended for changes in expected user behavior and predicted data throughput. Thus, the remaining bandwidth is increased if more interactive customers are expected, and reduced if the number remains or decreases. As less traffic is seen in other regions, the reservation is also reduced if it is expected that more connections can be moved immediately after. In the opposite case, the reservation increases accordingly.

From the above description regarding the assistance of the service (agent) of the present invention, it is possible to operate a line connection between the network provider and the end user (client) at an optimal cost / usage range, and thus with the network provider. It can be clearly seen that all of the end-users can save a significant amount of operating costs for maintaining the line connection.

8 is a block diagram showing a preferred embodiment of the second optimization step of the present invention. As already mentioned above, a URL list (pool) is used by the scheduler to set up a new acquisition and analysis process. The analysis process and the scheduler operate independently of each other in this regard. They communicate only indirectly via the URL list, where the scheduler gets the URL of the document to be requested whenever surplus bandwidth is available. New addresses are constantly added by the document parsing process, but these do not cause any direct interaction with the scheduler or direct processing of additional addresses. On the contrary, if the condition of the surplus bandwidth is met, the scheduler only starts a new acquisition process. If the scheduler is not in a state to request the document at a good time before completing the task, it is deleted from the URL list and completes the task. The URL list is preferably organized into a First-In-First-Out (FIFO) list in order to prioritize the address obtained first in time over another address.

9 conceptually illustrates the time course of an acquisition and analysis process by a second optimization step of the present invention. After document acquisition is completed, the document is analyzed for all references to other documents as references contained in the document, and if such a reference exists, the identified URL is placed in the URL pool. In the case of available bandwidth, an additional child process (thread) is now created by the URL, which in turn consists of a document acquisition and analysis process. The generation of the child process then ends, as described above, in accordance with the search depth defined by the client.

Fig. 10 shows a preferred form of performance of the JobScheduler in the object-oriented implementation of the second optimization stage according to the present invention. A job (document acquisition job) requesting a specific document via a URL is directly converted into an object of a 'job' class and transmitted to the job scheduler. If the job is a search query (a document search job), several search engines are referred to asynchronously with the agent in the program thread in charge of the query, and the returned address is entered in the URL list. Then, the standard procedure for obtaining the document is followed, that is, an object of class 'job' with the address found is created and sent to the job scheduler object. Tasks are sorted and processed according to their priority (eg, task completion time or provisional cost class). After completing the task, the task scheduler object notifies the WebAgent object of the completed task. The web agent object may then notify the user of the completion of the task, for example via e-mail, or send the data to the user using FTP upload.

In order not to burden the system with permanent status interrogation (polling), the functional components of the service according to the invention, in particular the task scheduler, are event-controlled. Thus, the task scheduler generally knows the current state of related processes (objects of the 'working' class) in response to a wake up signal sent for a new task or sent as the document acquisition ends. It only requires the related processes to report back. For the most urgent task-if enough bandwidth exists-the scheduler allows the start of one or more program threads. To do this, an object of class 'thread' is created for each program thread. Each of these objects, which initially have no properties, is linked to a Job object via an object of class Job: ThreadData. This facilitates access to the job data. The URL is also placed in an object of the 'WebRequest' class.

The first action of the new program thread is to get the referenced document through the 'Retriever' class. To communicate via TCP / IP (Transmission Control Protocol / Internet Protocol), this class creates an object of the 'Regulated Transceiver' class. The extractor object returns an object of class 'HTTPDocumentInfo' or 'FTPDocumentInfo', depending on the data of the document and the service protocol used in each case. If a document loaded via the HTTP protocol (hypertext transfer protocol) is involved, it is also possible to return to the type of medium. In this case, the program thread checks whether an HTML document is involved and creates an object of class "HTMLDocument" using the document data. This class provides an opportunity to search through the document for hyperlinks. The thread object checks all referenced documents for their attributes and determines whether to add the referenced documents to a list of already existing URLs. Selection criteria, such as size and file type, are the user's option. If the program thread terminates, then it sends the task scheduler object as a weather signal. If the URL list of the task is blank, the task is complete and the task scheduler deletes the task from the list of active tasks.

Fig. 11 shows an internal sequence when the task scheduler receives the weather signal 1 once. The task scheduler secures external access to all data structures via semaphores (2). This is necessary because, for example, the task scheduler can be addressed from another program thread, such as the 'webagent' class or the 'work' class object. After gaining security access, query the tasks to be processed about their status, report all completed tasks to the Web Agent object (A4, A4.1), or if sufficient bandwidth is available (B4) Allow a new program thread to start (B5, B5.1). Finally, the semaphore is released again (6) and waits for a new weather signal (1).

An embodiment of the interactive process is shown in FIG. 12 to show the sequence of the job process according to the present invention. The user (client) first transmits a job and appropriate job options to the agent (1). The agent creates a work object from it (1.1) and informs the scheduler of this object (1.2). The scheduler then asks the work object whether to start a program thread for document acquisition (according to the bandwidth available in each case) (1.2.1., 1.2.2, 1.2.3 and later 1.2. 5). The task starts these program threads (1.2.1.1., 1.2.2.1, 1.2.3.1, 1.2.5.1) and has no control over them. The program thread briefly reports to the scheduler before termination and notifies with completion (1.2.4, 1.2.6, 1.2.7, 1.2.8). The scheduler recognizes that there are no more documents to load, and thus declares (1.3) the completion of the task to the agent and returns to the waiting state. The agent starts a program thread (1.4) and informs the user that the work is complete via a mail object (1.4.1) (1.4.1.1). Thus, the agent is instructed to delete the job. By this deletion (2.1), the work object is discarded at the same time (2.2).

According to the present invention, on the one hand it is possible to optimize the access network to the information network in terms of load distribution, on the other hand, the Internet coupling between one network provider, intranet or end user and the other side of the Internet. Will be used as consistently as possible. In addition, optimal resource utilization can be seen between the process executed when the user's request is processed and the communication means required to send the job to the Internet.

Claims (38)

In an information network having at least one network node for supplying information, the information is searched and acquired according to a user's request, and the information is transferred to the user. 16. A method for conveying, comprising: first forwarding a request of said user to an agent, said agent executing said request in said information network separately from said user. The method of claim 1, wherein the requested information cannot be pre-positioned in the information network. 3. The method of claim 1 or 2, wherein the user obtains access to the information network by a network provider, and the agent is disposed with the network provider. The method of claim 1 or 2, wherein the user communicates directly with the information network by a coupling device, and wherein the agent is implemented as an autonomous service in the information network. 3. The system of claim 1 or 2, wherein the user is established within a company-internal network, wherein the network inside the company is connected to the information network. A method arranged at a connection interface between a network and the information network. 6. The method of any one of claims 1 to 5, wherein the agent executes the request at a time when network usage of the information network is low. 7. The method of claim 6, wherein prior to transmitting the information, the agent reserves bandwidth for additional interactive information traffic in the information network. 8. The method of any one of claims 1 to 7, wherein the information supplied by the agent is compressed before being sent to the user. The method of claim 1, wherein the information supplied by the agent is filtered by a request profile, and only the filtered information is sent to the user. 10. The method of any one of claims 1 to 9, wherein the user communicates the request to the agent by form. The agent according to any one of the preceding claims, in particular for use on the World Wide Web, which uses a hyperlink for archiving documents of the information network. And modifying the hyperlink usable in the case of information conveyed to the user, in order to refer to a document existing locally to the user. 12. The method according to any one of claims 1 to 11, wherein at least two search jobs are performed on the user's request, in particular for use in the world wide web where at least two search engines are provided. How to migrate to more than one different search engine. 13. The method of claim 12, wherein the information obtained for the user request is transferred by the search engine to the agent in parallel. Searching for information requested by the user and conveying the obtained information to the user through an information network having at least one network node for supplying the information. A device for performing the following steps: at least one agent for executing the user request in the information network, first communication means between the user and the agent, second communication means between the agent and the information network, Storage means for storing the user request transferred to the agent via the first communication means and the information obtained by the agent via the second communication means and separated from the user via the second communication means Process means for executing the user request. Information of the network information retrieval and transfer device. 15. The apparatus of claim 14, further comprising means for transferring the obtained information from the agent to the user in accordance with the user request. 16. The apparatus of claim 14 or 15, wherein the user obtains access to the information network by a network provider, and the agent is disposed with the network provider. 16. The information according to claim 14 or 15, wherein the user communicates directly with the information network by means of a coupling device, and wherein the agent is information of the information network implemented as an autonomous service in the information network. Search and transport device. 16. The system of claim 14 or 15, wherein the user is established within a company-internal network, wherein the network inside the company is connected to the information network. An information retrieval and transfer device of an information network arranged at a connection interface between a network and the information network. 19. An information retrieval and transfer apparatus according to any one of claims 14 to 18, wherein said agent comprises means for optimizing bandwidth for the transmission of information according to the transmission capacity present in each case. 20. The apparatus of claim 19, wherein the agent comprises means for reserving bandwidth for additional interactive information traffic in the information network. 21. Apparatus according to any one of claims 14 to 20, wherein the agent comprises means for compressing information to be sent to the user. 22. The method according to any one of claims 14 to 21, wherein the agent further comprises means for filtering information supplied by the agent, wherein information sent to the user is filtered by means for filtering according to a request profile. Apparatus for retrieving and transferring information in an information network. 23. The apparatus according to any one of claims 14 to 22, further comprising a form for transmitting the user request from the user to the agent. 24. A user as claimed in any one of claims 14 to 23, particularly for use on the World Wide Web, which uses a hyperlink for archiving documents of an information network. And means for modifying said hyperlink usable in the case of information conveyed to said user, for the purpose of referencing a document that exists locally. 25. Information retrieval and transfer of an information network as claimed in any one of claims 14 to 24, further comprising an operating system of the agent that provides means for parallel retrieval and acquisition of information in the information network. Device. The information is searched and acquired according to a user's request in an information network, and the information is transmitted through a line network having a transmission bandwidth that is variable in time. A method for conveying information to said information network comprising at least one network node for supplying information, said information comprising in each case at least one reference to additional information. Requesting information according to the user's request, analyzing information already received for a reference, filing a reference found in the analysis, and requesting the corresponding referenced information separately from the analysis process. Method comprising the steps. 27. The method of claim 26, comprising filing a reference found in the analysis into a list of URLs. 28. The method of claim 26 or 27, further comprising monitoring the transmission bandwidth of the circuit network and requesting the referenced information when there is available bandwidth in each case. 29. The method of any of claims 26 to 28, further comprising processing the filed reference according to a priority chain. 30. The method of any one of claims 26 to 29, further comprising creating a new program thread in each case, if there is available bandwidth. Searching for information requested by the user through an information network, and transferring the obtained information to the user via a line network having a transmission bandwidth that is variable in time. apparatus for conveying, wherein the information network comprises at least one network node for supplying information, the information comprising in each case at least one reference to additional information. (A) a-1) process user requests, and accordingly request information from the information network, A-2) analyze already acquired information for reference and A-3) process means for requesting the referenced information according to the result of the analysis, (B) storage means for storing said user request, already obtained information and a reference found in said analysis; and (C) information retrieval of an information network comprising time scheduling means for monitoring said transmission bandwidth of said circuit network and activating said process means for requesting referenced information according to said respective transmission bandwidths; And conveying device. 32. The apparatus of claim 31, wherein the process means for requesting the referenced information creates a new program thread for the request in accordance with the available transmission bandwidth. 33. A device according to claim 31 or 32, wherein a list of suitable URLs is provided as a storage means for storing the references found in said analysis. 34. An information retrieval and transfer device according to any one of claims 31 to 33, wherein a FIFO storage device is provided as a storage means for storing a reference found in said analysis. 35. An information retrieval and transport apparatus according to any one of claims 31 to 34, wherein additional information about an acquisition depth of said retrieval is filed in said storage means. 36. An information retrieval and transfer apparatus according to any one of claims 31 to 35, wherein all user requests are filed in said storage means. 37. An information retrieval and transfer apparatus for an information network according to any one of claims 31 to 36, wherein filtering conditions for retrieval defined by said user are filed in said storage means. 38. The method according to any one of claims 31 to 37, wherein the process means for requesting the referenced information compares the information of the coordination, which processed the request with other information, by means of the inherent nature of the information. Information retrieval and transport apparatus in an information network comprising means for prioritizing the priorities.