Advice-related systems – improving user interaction

Abstract

The present invention relates to improving system-user interaction related to advice, and in particular to allocation suggestions that assign specific values ​​to multiple physical entities and to any virtual entities.

Description

Technical Field

[0001] The present invention relates to the technical field of computer network processing technology on large data streams. Thereby, specific user preferences are received, and distribution proposals for a plurality of physical entities and any virtual entities are created based on the data stream, user preferences, and other factors.

Background Art

[0002] The applicant has noticed that there is no computer-implemented system that can allocate a limited amount of a plurality of physical and / or virtual entities. The allocation takes into account information from multiple information sources, preferences prompted by the user (which includes, inter alia, the evaluation of information sources or user ratings), and factors that at least define geographical information and application field information for each of the entities. The proposed allocation is compared with one or more comparison configurations based on user preferences and factors and provided to the user based on the comparison.

[0003] U.S. Patent No. 1,3935,198 describes a system for dynamically managing a supply chain by estimating the impact on the supply chain (logistical and / or financial) considering the potential to affect the vent and analyzing external information sources such as news channels. However, this is a strict system that focuses entirely on the supply chain and cannot be quickly changed to suit other needs or specific user preferences.

[0004] Other similar systems as described in U.S. Patent No. 7,680,719 are known in the art, but lack user interaction (either in that a user without the administrative authority to change the internal operation of the system cannot provide personalized preferences, or the information sources used are not properly screened, or there is no "practical" filter to exclude proposals that are, for example, geographically impossible or highly unlikely).

[0005] The present invention aims to address at least some of the above-mentioned problems. [Prior art documents] [Patent Documents]

[0006] [Patent Document 1] U.S. Patent No. 1,3935,198 [Patent Document 2] U.S. Patent No. 7,680,719 [Overview of the project]

[0007] In a first aspect, the present invention provides the method according to claim 1.

[0008] In a second aspect, the present invention provides the system described in claim 14.

[0009] In a further embodiment, the present invention provides the use described in claim 15.

[0010] Further preferred embodiments and their advantages are described in the detailed description and the claims. [Modes for carrying out the invention]

[0011] As used herein, the following terms have the following meanings:

[0012] As used herein, "one" refers to both singular and plural objects unless the context explicitly indicates otherwise. For example, "one section" refers to one or more sections.

[0013] As used herein, “includes,” “equips,” and “has” are synonymous and are inclusive or non-exclusive terms, for example, specifying the presence of a component followed by another component, and excluding or not excluding the presence of additional, non-enumerated components, features, components, elements, or processes known in the art or disclosed herein.

[0014] The enumeration of numerical ranges by endpoint includes all numbers and fractions contained within that range, as well as the enumerated endpoints themselves.

[0015] The term "information source" can relate to any data channel, data stream, dataset, or data subscription service that provides content in file-based, update-based, synchronization-based, and / or message-based formats, thereby each information source that does not provide data within a pure message can be considered equivalent to a message base, thereby the updated portion of the data can relate to one or more new messages. Messages may be provided on demand by requesting them from storage, or they may be sent to the user in real time, for example, as email, push message notifications, in-app messages, a dedicated web interface, a web dashboard, or news items in a feed. Messages may be delivered within the context of an account held by the subscription and / or provider of the information source.

[0016] Unless otherwise defined, all terms used in disclosing this invention, including technical and scientific terms, are intended to be understood by those skilled in the art to which this invention pertains. Furthermore, definitions of terms are included to better understand the teachings of this invention.

[0017] In a first aspect, the present invention relates to a computer implementation method for providing a user with a system-generated executable notification based on the interaction between the user and the system, and includes the following steps: a. Encourage the user to provide the system with multiple user preference values. b. The server repeatedly receives multiple messages originating from multiple sources, and these messages are directly and / or indirectly associated with one or more predefined physical entities and one or more predefined virtual entities, thereby storing the received messages as context data. c. Assign at least one factor from each of at least two factor sets to each of the physical entities and each of the virtual entities, such that the first at least two factor sets relate to the geographical evaluation of the physical and virtual entities, and the second at least two factor sets relate to the application areas of the physical and virtual entities. d. An allocation proposal is created by distributing values ​​to multiple physical and virtual entities, thereby making the aggregate allocation value of the physical entities non-zero, and thereby making the aggregate allocation value of the physical and virtual entities equal to a user-specified value, and the allocation proposal includes physical and virtual entities with non-zero allocation values ​​and each of the associated non-zero allocation values. e. Provide the user with actionable notifications, which include allocation proposals. At least one user preference is characterized by defining the user's evaluation of each information source, Thus, the allocation value is calculated based on user preferences, factors, and the values ​​of the information source message; thereby, the allocation proposal is compared with a comparison configuration that includes multiple physical and virtual entities having non-zero allocation values; thereby, the non-zero aggregate allocation values ​​of the entities in the comparison configuration are equal to user-defined values; thereby, the allocation values ​​of the comparison configuration are based on user preferences and factors; thereby, actionable notifications are provided based on the comparison between the allocation proposal and the comparison configuration.

[0018] The methods defined above provide advantageous solutions for entity management in numerous fields. Given the diversity of the defined methods, entities can vary greatly in their definition, ranging from a wide variety of tangible tradable goods such as foodstuffs, automobiles, raw materials, general supplies, coupons, art, possessions, and livestock, to more cumbersome tradable goods such as stocks, common stocks, bonds, cash, real estate, valuables, and investment funds. The allocation of values ​​for such entities can, for example, be assigned to the location, owner, price, cost, characteristics of the storage device, and the transmission of the entity. The optimized allocation of such parameters / values ​​depends heavily on many agents, some explicitly, others implicitly or indirectly. The first agent can be a user preference, which often has a strongly deterministic nature in allocation and can act as a guide for other agents to perform optimization, particularly as a user preference, in this case reflecting the use of at least some other agents. The second agent is multiple information sources. As already mentioned, user preferences, in particular, define the user's evaluation of information sources, which guides the allocation process, and the evaluation determines the "weight" each information source has in the allocation process. A user may prefer a particular information source to others (for example, Twitter feeding official Dow Jones news), and there may be lower evaluations that further specify the weights for specific subsets of information from some information sources. Thirdly, an agent can define a factor set as a set of so-called "factors" (elements). Factors are assigned to each entity and define specific characteristics of that entity. At least one of the factors defines the entity's geographical information, and at least one factor defines the entity's field of application. Geographical information is typically the country, region, or continent that represents the entity (however, it can be a city, state, or one or more of the aforementioned groups). For example, in the case of a physical object such as raw materials, it could be the place of origin (manufacturing area or something similar) or the actual physical location. In the case of a virtual object, it could represent the location of its owner or the location of the related physical entity (e.g., in the case of shares, the location or area of ​​the company issuing the shares). Regarding the field of application, it can reflect the nature of the entity (e.g., shares, raw materials, possessions, food, medical devices, etc., as defined above), or more generally, the technological field to which healthcare, technology, etc., belong. It has often been found that such factors will have a strong influence on the optimal allocation. For example, in the allocation of entities such as food products, geographical location would be important as a "category" (meat, dairy products, grains, storable goods, etc.) for optimal allocation. The same applies to constructing investment portfolios, for example, where location can affect the physical safety of goods and therefore their reliability / value, or in the case of stocks, their volatility, so it is important that entities (physical or virtual) are defined by such factors. Furthermore, the entity's factors define the basis that allows a computer system to establish associations with specific messages that affect the entity, which simplifies the calculation of the entity's allocation value. In these cases, user-defined constraints often have to be imposed on the allocation values, regardless of whether they are due to limited available cargo space (weight and / or volume), limited refrigeration means, manpower, and / or even budgetary constraints. Therefore, these are set as the maximum value of the aggregated allocation values for the entity. Finally, the system generates an allocation proposal that basically includes all lists of entities with non-zero allocation values and the associated allocation values of the entities. This is provided to (qualified) persons and approved persons, and can be further analyzed, regardless of whether there are changes or corrections (for example, if the person knows information that was not yet in the system, or if there are simply certain drawbacks based on previous experience). However, before the allocation proposal is provided to the user, it is compared with one or more other allocation proposals to verify its effectiveness / efficiency. These "known" comparison configurations can be obtained from configurations used in the past (configurations generated in old systems, both the plans generated by the user and external information, so-called benchmark allocations). To meet the requirements provided to the user, the allocation proposal must "exceed" the comparison configuration based on specific criteria. Finally, it can be understood that the allocation proposal is provided as an actionable notification and can be executed (or changed by the user) to actually affect the (technical) effect.

[0019] In this specification, the terms "value" or "allocation value" can be understood to represent specific information regarding an entity. These values can be the quantity of the entity proposed to be "used" such that it is transferred, purchased, and stored.

[0020] None of the prior art systems enable such complex interactions between the user and the system to generate distribution proposals. For example, the system will generally prevent the user from providing an evaluation of the input information. However, it must be understood that human experience can correctly evaluate specific information sources depending on the subject matter and guide the system (e.g., when distributing emergency supplies to address a humanitarian crisis, by placing more weight on certain reliable accounts rather than classical sources such as newspapers that cannot deliver emergency content at the same speed, the user may place a high value on "fast" social media sources like Twitter). Conversely, more volatile news sources (such as typical self-published sources like Twitter, Facebook, or self-published web content) can be rated lower than traditional news sources or other more regulated channels when time is not an issue.

[0021] In a preferred embodiment, the user preferences include at least a regularization type preference and a regularization amount preference. Preferably, thereby, the regularization type preference is related to either regularization without variable selection or regularization with variable selection. More preferably, the regularization without variable selection is related to Tikhonov regularization or ridge type regularization. More preferably, the regularization with variable selection is related to Lasso regularization. Furthermore, preferably, the regularization amount preference is related to a value indicating the strength of regularization, more preferably a real value.

[0022] Here, "regularization" is a mechanism for evaluating the messages of an information source, taking into account the uncertainty and / or importance of the information source that provides the messages that influence the output of the evaluation. This mechanism addresses the problem that even though messages are always generated with a high level of "noise"—that is, with high uncertainty—the majority of the messages contribute little (if any) to the final value produced by the evaluation. Thus, regularization reflects the level of distrust regarding the output produced by the evaluation, based on a stable mathematical concept. Therefore, regularization can be understood as a three-step approach. Define several normal concepts for each information source, typically the absence of a message, or a "zero signal." The message stream from all sources is modeled as the sum of content and noise, or "real signal + noise". To bring the system closer to normal, noise is removed by distributing the evaluation output or the "overall signal".

[0023] In one modified example, the regularization is reduced to automatically adjust the impact of incoming messages according to uncertainty, while retaining all messages from all sources, preferably updating the evaluation threshold for each arrival. The notification trigger can then be related to messages from any source, thereby allowing the regularization to suppress some anomalies or outliers. This allows the user to control the strength of the regularization, preferably through a regularization quantity preference. In corresponding and / or related preferred embodiments, the regularization type preference relates to regularization without variable selection, which is preferably related to Tychonov regularization or Ridge regularization, and the regularization quantity preference relates to a value, preferably a real number, indicating the strength of the regularization.

[0024] In another variation, regularization corresponds to automatically suppressing, over a period of time, any information source deemed too uncertain or too insignificant, while adjusting other messages according to uncertainty. The allocation proposal can then be based only on messages from unsuppressed information sources, thereby allowing regularization to further adjust the impact of unsuppressed messages. This allows the user to control the intensity of regularization, preferably through a regularization quantity preference. In corresponding and / or related preferred embodiments, the regularization type preference relates to regularization using variable selection, which is preferably related to Lasso regularization, and the regularization quantity preference relates to a value, preferably a real number, indicating the intensity of regularization.

[0025] Regularization provides users with even more useful tools for evaluating values ​​from various sources, which greatly assists in the task of allocating values ​​to entities.

[0026] In a preferred embodiment, the values ​​allocated to entities in the allocation proposal are further calculated based on an optimization model selected by at least one user, the optimization model being selected from a list including at least the Black-Litterman model, the Markowitz model, and the minimum variance model.

[0027] In a preferred embodiment, the information sources include direct contextual information sources, structured peripheral information sources, unstructured peripheral information sources, and speculative direct contextual information sources, thereby enhancing the data of the direct contextual information sources through characteristic engineering, and preferably thereby enhancing the data of the speculative direct contextual information sources and / or structured peripheral information sources through characteristic engineering.

[0028] In a preferred embodiment, the multiple information sources include at least one unstructured peripheral information source. At least one message originating from an unstructured peripheral source contains a raw text string. To evaluate the relevance of a message to contextual data, the process includes a step of comparing the message to the contextual data using natural language processing of the raw text string. This means the allocation value is calculated based on the evaluation.

[0029] In a further preferred embodiment, messages from unstructured peripheral sources are processed before the use of calculated allocation values, thereby the processing includes structuring the messages from unstructured peripheral sources via a self-learning artificial intelligence, thereby the AI ​​operates under user-adjustable hyperparameters.

[0030] In a preferred embodiment, physical entities and virtual entities are each assigned a predictability rating based on at least historical information about the physical and virtual entities in the context data, thereby the allocation value of the allocation proposal is calculated based further on the predictability rating.

[0031] Knowledge of the "predictability" of a particular entity can help predict further development and can be useful in creating a certain harmony or safety in the overall allocation proposal.

[0032] In a preferred embodiment, the prompted user preference includes one or more constraints, the constraints including at least a user-defined minimum predictability rating for physical and / or virtual entities having non-zero allocation values ​​in the allocation proposal, and / or a user-defined minimum aggregated predictability rating for physical and virtual entities having non-zero allocation values ​​in the allocation proposal.

[0033] As mentioned above, ignoring the predictability rating of entities can lead to highly risky allocation proposals, which may be efficient or advantageous, but only under the assumption that everything proceeds according to plan. Even with a reasonable stance on the future impact of each entity, there are still significant differences in the potential variability of each entity's impact (e.g., catastrophic harvests, production problems, etc., will have a stronger impact on certain entities), and this is not reflected in the predictions. Therefore, depending on the circumstances (urgency, location, etc.), users may enable higher or lower predictability ratings for allocation proposals.

[0034] In a preferred embodiment, the prompted user preferences include one or more constraints for creating an allocation proposal, the constraints including at least a maximum and / or minimum number of physical and / or virtual entities having non-zero allocation values ​​in the allocation proposal, and / or maximum and / or minimum allocation values ​​for physical and / or virtual entities in the allocation proposal.

[0035] In a preferred embodiment, user preference includes a historical allocation configuration comprising one or more physical and / or virtual entities having non-zero historical values. This allows the system to define a target cost function based on contextual data adapted to calculate the rebalancing costs required to transform past allocation configurations into allocation proposals.

[0036] Past allocation configurations include the "current" or "most recent" (or expected) allocation configuration and represent the last known value for an entity (or at least an entity with a non-zero allocation value in any of the allocation proposals and / or past allocation configurations). In some cases, changing a value will result in certain costs, administrative, governmental, or operational, such as certain taxes or transportation costs. Note that these costs do not necessarily have to be "monetary" and could be the cost of time, labor, or space (volume / weight) required to implement the particular change. These costs must be considered by the user in order to present an allocation proposal, as they may affect the merits of the proposed allocation proposal (for example, excessively high time costs may negate all other merits of the proposal).

[0037] In a preferred embodiment, the prompted user preferences include at least one constraint for generating allocation proposals, the constraint being the maximum rebalancing cost for converting past allocation configurations into allocation proposals.

[0038] In a preferred embodiment, contextual data is further processed through ensemble learning. Combining multiple learning algorithms improves the overall predictive preference of the process.

[0039] In a preferred embodiment, user preference includes at least one constraint on context data based on the calculation of an allocation value, thereby the constraint depends on at least one of the period over which messages containing context data are generated and the user's evaluation of the sources of information for messages containing context data. The user may, for example, focus on information for a specific period to simplify the evaluation process.

[0040] In a second aspect, the present invention relates to a computer system that generates executable notifications to a user based on the interaction between the user and the system, and the present invention relates to a computer system that generates executable notifications to a user based on the interaction between the user and the system. a. A server comprising a processor, tangible non-volatile memory, program code residing in memory for issuing commands to the processor, and connection means for connecting a user's device to one or more remote servers, b. A user's device, the user's device including a processor, tangible non-volatile memory, program code residing in memory for issuing commands to the processor, a screen for displaying information to the user, preferably input means for receiving user input means from the user, and connection means for connecting to a server via a computer network. c. One or more remote servers, each remote server being associated with at least one of a plurality of information sources, and each one or more remote servers including connection means for connecting to the server via a computer network, d. At least one computer-readable medium, the at least one computer-readable medium being accessible to a server and including a database, the database including context data and user preference data, and preferably the at least one computer-readable medium being contained within the server, Includes, A computer system is configured to generate executable notifications to a user, and generating executable notifications is: A step of prompting the user to provide values ​​for multiple user preferences to the system, A server repeatedly receives multiple messages originating from multiple sources, wherein the multiple messages are directly and / or indirectly related to one or more predetermined physical entities and one or more predetermined virtual entities, and the received messages are stored as context data. A step of assigning at least one factor from each of at least two factor sets to each physical entity and each virtual entity, wherein the first at least two factor sets relate to the geographical evaluation of the physical and virtual entities, and the second at least two factor sets relate to the application areas of the physical and virtual entities. A step of creating an allocation proposal by distributing values ​​to multiple physical entities and virtual entities, wherein the aggregate allocation value of the physical entities is non-zero, the aggregate allocation value of the physical entities and virtual entities is equal to a user-specified value, and the allocation proposal includes physical entities and virtual entities having non-zero allocation values, and each of the associated non-zero allocation values. A step of providing an actionable notification to the user, wherein the actionable notification includes an allocation proposal, Includes, This means that at least one of the user preferences defines the user's evaluation of each information source. The allocation value is calculated based on user preferences, factors, and message values ​​from the information source. The allocation proposal is compared to a comparison configuration that includes multiple physical and virtual entities with non-zero allocation values. The non-zero aggregate allocation values ​​of the entities in the comparison configuration are equal to user-specified values, and the allocation values ​​of the comparison configuration are based on user preferences and factors.

[0041] In a third aspect, the present invention relates to the use of the method according to the present invention in a system according to the present invention.

[0042] The present invention can be further illustrated by the following non-limiting examples, which are not intended to limit the scope of the invention, nor should they be construed as limiting the scope of the invention. [Examples] [Examples]

[0043] <Physical Entity Collection (Collectibles)> In this example, the invention relates to monitoring by a user of a physical object collection, which includes multiple physical goods or physical objects stored or located in multiple locations. Physical objects can be valuables with intrinsic value, such as gold, diamonds, (older) automobiles, wine, whiskey, fine art paintings, musical instruments, or jewelry, but can also relate to valuables whose value is related to a right granted to an owner, such as bearer bonds. In yet another example, physical objects can relate to real estate, for example, objects can correspond to houses, buildings, apartments, apartment blocks, garages, or garage blocks. In a preferred exemplary embodiment, each object corresponds to a physical entity record stored in a database, and vice versa. In one embodiment, physical objects can relate to perishable articles that may be damaged by accidents related to excessive temperature or high moisture content, such as automobiles, wine, whiskey, fine art paintings, or musical instruments. Here, the first allocation proposal can relate to maximizing the total number of objects across multiple locations that remain intact over a given time period, for example, one year. Such allocation proposals can generate notifications whenever any location is likely to suffer damage, based on messages originating from one of the sources, taking into account different location measurements and predictions available from different sources. However, a second allocation proposal can relate to maximizing the chances that at least one object remains intact over a given period of time, for example, one year. This could relate, for example, to a hard disk drive or flash drive containing highly confidential information that has been duplicated across multiple locations. Thereafter, the hard disk drive or flash drive may or may not be connected to a computer network, depending on the amount of confidentiality required, and thereafter, the data on the drive may or may not be encrypted.Such highly confidential information may relate to, for example, passwords, credentials, or cryptocurrency assets. In such allocation proposals, a prediction or measurement of a single location may lack relevance and therefore not require notification unless a significant portion of the objects are stored at that single location. Whether the latter is true can be indicated by a strong change in the output of the allocation proposal if an indicator of an incident is detected at that single location. In various relevant exemplary embodiments, one source of information may be an alarm service that provides alerts related to location and / or building integrity and security, such as alarms or incident detection at object locations. Another source of information may relate to a raw text service that provides news messages. This allows the detection of a location name and / or a building name associated with the location to indicate that the message is relevant enough to take the allocation proposal into consideration.

[0044] The advantage of this invention lies in the user's control over what is relevant and irrelevant, and to what extent notifications are needed. Regularization allows end-users to select information sources that are meaningful to them. For example, a news message from a single source regarding a fire at a location may be deemed irrelevant and may be considered an outlier based on regularization, while multiple messages with very similar content may be highly relevant and therefore trigger a notification. This allows the strength of regularization to set a threshold based on user preference. Furthermore, users can indicate whether they prefer regularization without variable selection or with variable selection. Overall, users are freed from the burden of manually examining all messages potentially relevant to their physical object collection. Regularization provides users with an even more useful tool for evaluating the value of various information sources. Therefore, this invention is valuable in deciding whether to continue subscribing to information sources (paid or not) and which subscriptions can be removed. [Examples]

[0045] <Portfolio of Tradable Products> In this example, the context data relates to all information relating to a portfolio containing multiple tradable goods related to at least two physical entities. The context data includes multiple data records, each containing at least two physical entity records, and each physical entity record relates to a physical entity. A physical entity can be any physical investment product, e.g., gold, diamonds, (older) cars, wine, whiskey, fine art paintings, collector's items, musical instruments, jewelry, bearer bonds, houses, buildings, apartments, apartment blocks, garages, garage blocks. Beyond physical entity records, the portfolio may or may not contain data records relating to entities that are not physical entities. Each data record is a set of structured inputs with numerous attributes. One attribute may be the type of entity, e.g., stock, common stock, bond, cash, real estate, valuables, investment fund, etc. Another attribute may indicate the quantity of the entity. Yet another attribute may relate to the start or end date of the agreement related to that entity. As in Example 1, another attribute may be the location of the entity, for example, with respect to a set of street addresses or GPS coordinates. Another attribute can enable a natural language description of an entity that can be considered through natural language processing.

[0046] The allocation proposal relates to the technical objectives of the portfolio. The composition of the allocation proposal relates to training multiple parameters historically for each information source. The allocation proposal may or may not be related to the application of Black-Litterman optimization. Information sources include prestructured data, including records with inputs in multiple fields, and unstructured data, including raw text. The prestructured data provides metrics such as news intensity, public opinion on the news, and the number of news readers, at both the company and aggregate levels. The prestructured data can be linked to an index. Users are further given the possibility to change the information sources on which the public opinion score is constructed to respond to developments in social media use and developments in official press conferences. Here, public opinion is typically measured in relation to several historical criteria, but the relevant criteria can be modified to place more emphasis on credibility with people who shape certain public opinion. The processing flow can also be modified, for example, so that date updates occur daily or more frequently.

[0047] Therefore, automated processing provides added value to the monitored portfolio, along with manual processing. In particular, the present invention provides a comprehensive alternative to purely manual processing, a partial alternative to lower-level decisions, and information generated in the processing. Significant changes in forecasts can trigger explanatory notifications sent to the user, e.g., portfolio manager, who can then rebalance the portfolio. The present invention further enables the identification of whether a particular entity and / or related financial instrument is under review. Furthermore, the user can be alerted to a large amount of negative news accumulating around a particular entity. Moreover, for example, if the shares of several companies plummet, indirect relationships involving multiple entities can be established based on the related companies and sectors that would be affected.

[0048] The method by which notifications can be generated for the user is further similar to that of Example 1. In this example, the user is also relieved of the burden of reviewing all messages that may be potentially relevant to their portfolio. Regularization further provides the user with a useful tool for evaluating the value of various information sources. Thus, the present invention is valuable in determining whether subscriptions to information sources (paid or not) should be continued and which subscriptions can be canceled. [Examples]

[0049] <Allocation of crisis supplies> In this example, contextual data relates to information about local crisis situations such as famine, disease outbreaks, and natural disasters. Sources in this case include traditional media (television, newspapers, online news) and social media (Twitter, Facebook, Instagram, Tencent, Weibo, YouTube). Entities in this case can relate to food, medicine, infrastructure supply (tents, etc.), financial assistance, etc. The user can then provide specific values ​​in response to prompts from the system, including an evaluation of the source. In these cases, unstructured sources may be highly relevant as they quickly identify dangerous situations at the local level (e.g., a specific village where an outbreak occurred), while larger media coverage might miss these sources or refuse to report them without further verification. Therefore, the user can give a greater evaluation to such sources. Other information that may be conveyed includes geographical details and the nature of the crisis. The system will access a large database of entities containing information on entities, such as grain supplies in surrounding areas (countries, regions, cities, etc.), medicines, medical personnel from specific organizations (Doctors Without Borders, etc.), and transport capacity (the number of cargo planes with a capacity of Y in country Z, etc.). Note that the database will be updated through the influx of information from the sources. Based on this, the system can generate time-based proposals for allocating resources to address a crisis. The proposals can allocate values ​​to multiple entities, indicating whether they will be used (for example, if a particular supply cannot be delivered from a specific location within a given time), how many of them will be used, and how they will be delivered.

[0050] The present invention is not limited to the embodiments described above, and it is assumed that modifications can be made to the presented embodiments without reconsidering the appended claims. For example, although the present invention has been described with reference to the allocation of both physical and virtual goods, products, etc., it is clear that the present invention can be used for other purposes.

Claims

1. A computer implementation method for providing a user with a notification of system generation based on the interaction between the user and the system, wherein the computer implementation method is a. Prompting the user's device to provide the user's device with values ​​for a plurality of user preferences for the system to the input means of the user's device, b. A step of a server repeatedly receiving multiple messages originating from multiple sources, wherein the multiple messages directly and / or indirectly relate to one or more predetermined physical entities and one or more predetermined virtual entities, and the information relating to the one or more predetermined physical entities and one or more predetermined virtual entities in the received messages includes context data, and the virtual entities are non-physical entities. c. A step of assigning, by the server, at least one factor from each of at least two sets of factors to each of the physical entities and each of the virtual entities, wherein the first set of at least two factors relates to the geographical assessment of the physical entities and the virtual entities, the second set of at least two factors relates to the application area of ​​the physical entities and the virtual entities, and the geographical assessment of the virtual entities indicates the location of the owner of the virtual entities or the location of the physical entities to which the virtual entities relate, d. A step of creating an allocation proposal by distributing values ​​to a plurality of physical entities and virtual entities by the server, wherein the distributed and aggregated values ​​to the physical entities are non-zero, the distributed and aggregated values ​​to the physical entities and virtual entities are equal to values ​​predefined by the user, and the allocation proposal includes the physical entities and virtual entities having non-zero distributed values, and the non-zero distributed values ​​associated with the physical entities and virtual entities, respectively. e. A step of providing a notification to the user by the server, wherein the notification includes the allocation proposal, Includes, At least one of the user preferences defines a user rating for each of the information sources, and the value for the user preference includes a value indicating the user rating for the information source. The allocated value is calculated based on the value for the user preference, the factor, and the message of the information source; the allocation proposal is compared with a comparison configuration comprising a plurality of physical and virtual entities having non-zero allocated values; the allocated and aggregated non-zero values ​​for the entities in the comparison configuration are equal to values ​​predefined by the user; the allocated values ​​for the comparison configuration are based on the value for the user preference and the factor; and the notification is provided based on the comparison between the allocation proposal and the comparison configuration. The user preference includes at least a regularization type preference and a regularization quantity preference, the value for the user preference includes a value indicating the type of regularization and / or the intensity of the regularization, and the regularization is Steps to define the normality of each information source, The step of modeling the message stream from all sources as the sum of content and noise, To bring it closer to normal, the noise is removed by outputting an evaluation output, Includes, The aforementioned regularization preference relates to either regularization without variable selection or regularization with variable selection. The aforementioned regularization without variable selection is related to Tychonov regularization or ridge regularization, The regularization involving the aforementioned variable selection is related to Lasso regularization, The preference for the amount of regularization is related to a value indicating the intensity of the regularization. The aforementioned message is retained from all sources, A computer implementation method characterized in that each message is given uncertainty and importance as attributes, and the regularization includes automatically suppressing any information source for which the uncertainty of the message is judged to be too high or the importance is too low for a predetermined period, and adjusting other messages other than the message of the suppressed information source according to the uncertainty.

2. The computer implementation method according to claim 1, characterized in that the values ​​allocated to the entities of the allocation proposal are further calculated based on at least one user-selected optimization model, the optimization model being selected from a list including at least the Black-Litterman model, the Markowitz model, and the minimum variance model.

3. The computer implementation method according to any one of claims 1 to 2, wherein the information source includes a direct contextual information source, a structured peripheral information source, an unstructured peripheral information source, and a speculative direct contextual information source.

4. The aforementioned multiple information sources include at least one unstructured peripheral information source, At least one message originating from the aforementioned unstructured peripheral information source includes a raw text string, To evaluate the relevance of the message to the context data, the step includes comparing the message to the context data using natural language processing of the raw text string, The aforementioned allocated values ​​are calculated based on the aforementioned evaluation. A computer implementation method according to any one of claims 1 to 3, characterized in that

5. The computer implementation method according to claim 3 or 4, wherein messages from the unstructured peripheral information source are processed before the calculated allocated values ​​are used.

6. The computer implementation method according to any one of claims 1 to 5, characterized in that the physical entity and the virtual entity are each assigned to a predictability rating based on at least past information of the physical entity and the virtual entity in the context data, and the allocated value of the allocation proposal is further calculated based on the predictability rating.

7. The computer implementation method according to any one of claims 1 to 6, characterized in that the prompted user preference includes one or more constraints, the constraint includes at least a user-defined minimum predictability rating based on at least historical information of the physical entities and / or virtual entities having non-zero allocated values ​​in the allocation proposal, and / or the constraint includes a user-defined minimum aggregated predictability rating based on at least historical information of the physical entities and virtual entities having non-zero allocated values ​​in the allocation proposal.

8. A computer implementation method according to any one of claims 1 to 7, characterized in that the prompted user preferences include one or more constraints for creating the allocation proposal, the constraints include at least the maximum and / or minimum number of physical entities and / or virtual entities having non-zero allocated values ​​in the allocation proposal, and / or the constraints include the maximum and / or minimum values ​​of the allocated values ​​of the physical entities and / or virtual entities in the allocation proposal.

9. The user preference includes a historical allocation configuration comprising one or more physical entities and / or virtual entities having non-zero historical values, Based on the context data adapted to calculate the rebalancing cost for converting the aforementioned past allocation configuration to the aforementioned allocation proposal, the objective cost function is defined in the system. A computer implementation method according to any one of claims 1 to 8, characterized in that

10. The computer implementation method according to claim 9, characterized in that the prompted user preference includes at least one constraint for creating the allocation proposal, the constraint being the maximum value of the rebalancing cost for converting the past allocation configuration to the allocation proposal.

11. The computer implementation method according to any one of claims 1 to 10, wherein the user preference includes at least one constraint on the context data based on the calculation of the allocated values, the constraint depending on at least one of the period during which the message containing the context data is generated and the user evaluation of the information source of the message containing the context data.

12. A computer system that generates notifications to a user, wherein the system is based on interaction between the user and the system, a. A server comprising a processor, tangible non-volatile memory, program code residing in the memory for issuing commands to the processor, and connection means for connecting the user's device to one or more remote servers, b. The user's device, the user's device including a processor, tangible non-volatile memory, program code residing in the memory for issuing commands to the processor, a screen for displaying information to the user, input means for receiving user input means from the user, and connection means for connecting to the server via a computer network, c. The remote server comprising one or more remote servers, each of which is associated with at least one of a plurality of information sources, and each of the one or more remote servers includes connection means for connecting to the server via the computer network, d. A computer-readable medium comprising at least one computer-readable medium, the at least one computer-readable medium being accessible to the server and including a database, the database including context data and user preference data, the at least one computer-readable medium including a computer-readable medium contained within the server, The computer system is configured to generate the notification to the user, and generating the notification is: The steps of prompting the user's device to provide the user's input means of the device with values ​​for a plurality of user preferences for the system, A server repeatedly receives multiple messages originating from multiple sources, wherein the multiple messages directly and / or indirectly relate to one or more predetermined physical entities and one or more predetermined virtual entities, and the information relating to the one or more predetermined physical entities and one or more predetermined virtual entities in the received messages includes context data, and the virtual entities are non-physical entities. A step of assigning, by the server, at least one factor from each of at least two sets of factors to each of the physical entities and each of the virtual entities, wherein the first set of at least two sets of factors relates to the geographical assessment of the physical entities and the virtual entities, the second set of at least two sets of factors relates to the application area of ​​the physical entities and the virtual entities, and the geographical assessment of the virtual entities indicates the location of the owner of the virtual entities or the location of the physical entities to which the virtual entities relate, A step of creating an allocation proposal by distributing values ​​to a plurality of physical entities and virtual entities by the server, wherein the distributed and aggregated values ​​to the physical entities are non-zero, the distributed and aggregated values ​​to the physical entities and virtual entities are equal to values ​​predefined by the user, and the allocation proposal includes the physical entities and virtual entities having non-zero distributed values, and the non-zero distributed values ​​associated with the physical entities and virtual entities, respectively. A step of providing a notification to the user by the server, wherein the notification includes the allocation proposal, Includes, At least one of the user preferences defines a user rating of each of the information sources, and the value for the user preference includes a value indicating the user rating of the information source. The allocated value is calculated based on the value for user preference, the factor, and the message of the information source; the allocation proposal is compared to a comparison configuration comprising a plurality of physical and virtual entities having non-zero allocated values; the allocated and aggregated non-zero values ​​for the entities in the comparison configuration are equal to values ​​predefined by the user; the allocated values ​​for the comparison configuration are based on the value for user preference and the factor; and the notification is provided based on the comparison between the allocation proposal and the comparison configuration. The user preference includes at least regularized preference and regularized quantity preference, and regularization is, Steps to define the normality of each information source, The step of modeling the message stream from all sources as the sum of content and noise, To bring it closer to normal, the noise is removed by outputting an evaluation output, Includes, The regularization preference relates to either regularization without variable selection or regularization with variable selection, and the value for the user preference includes a value indicating the type of regularization and / or the intensity of the regularization. The aforementioned regularization without variable selection is related to Tychonov regularization or ridge regularization, The regularization involving the aforementioned variable selection is related to Lasso regularization, The preference for the amount of regularization is related to a value indicating the intensity of the regularization. The aforementioned message is retained from all sources, A computer system characterized in that each message is given uncertainty and importance as attributes, and the regularization includes automatically suppressing any information source for which the uncertainty of the message is judged to be too high or the importance is too low for a predetermined period, and adjusting other messages other than the message of the suppressed information source according to the uncertainty.

13. A computer system that performs the method according to any one of claims 1 to 11, in the system described in claim 12.