A method for determining a sub-set of listed items

Abstract

A method (300) for determining a sub-set of listed items (116) from a main set of listed items held on a central server (104) communicatively connected to a client computer (102) is disclosed. The listed items pertain to products and / or services, each of them being associated with different attributes. The method comprises obtaining (302) by the central sever (104) an initiation request (110) from the client computer (102), in response to the initiation request (110), transmitting (304) a user input request (112) from the central server (104) to the client computer (102), in response to the user input request (112), obtaining (306), at the central server (104), user input (114) retrieved at least partly via the client computer (102), ranking (306), at the central server (104), the listed items of the main set by using a machine learning model (414) based on the user input (114) obtained and the different attributes associated with the products and / or services pertaining to the listed items, assigning (308), at the central server (104), a number of top ranked listed items as the sub-set of listed items (116), and transmitting (310) the sub-set of listed items (116) from the central server (104) to the client computer (102).

Description

[0001] A METHOD FOR DETERMINING A SUB-SET OF LISTED ITEMS

[0002] Technical Field

[0003] The disclosure generally relates to methods and apparatuses using a machine learning model for efficiently selecting items in a digital market place that matches a user’s needs or wishes without compromising the user’s integrity. More particularly, it is disclosed herein a method for determining a sub-set of listed items from a main set of listed items held on a central server communicatively connected to a client computer, a method for obtaining the sub-set of listed items, and apparatuses and computer program products related thereto.

[0004] Background Art

[0005] Today, when selecting among items listed in a digital market place it is common practice to use filters, keyword search and other similar features for identifying a subset of listed items that fit the needs or wishes of a user. This is made possible by having the different products and services linked to the listed items associated with different attributes. For instance, a pair of training shorts may be linked to a gender, e.g. male, female or universal, a color, a material type, a brand, a collection, a price, availability, etc. By using the filter functionality, it is made possible for the user to sort out irrelevant items such that a remaining list of items comprises the subset of listed items that matches filter settings. In addition or as a complement to using the filter functionality, it is possible to find relevant products by using different keywords. In a simple form, the keywords input is compared to the attributes of the listed items, and if there is a match, the items are considered to be among the subset of listed items that meet the conditions associated with the input keywords. In a more advanced form of keyword search, the input keywords are combined with learnings made from previously made keyword searches. For instance, a keyword search engine, often based on a neural network or similar, may take into account learnings made from similar searches being performed in the past. Listed items that caught the interest of previous users making similar searches can for instance be decided to form part of the subset of listed items.

[0006] In addition to using filters and keyword searches, it is today common practice for digital market places to continuously retrieve data about user activity and based on this data provide recommendations, that is, the subsets of listed items may also be determined based on what the user in the past has been interested in and also purchased. By continuously learning the user’s preferences, it is possible over time to provide more relevant recommendations. Further, by comparing traces developed over time from one user with traces from other users, also these traces developed over time, it is made possible to link different user groups together in a precise manner, in turn resulting in that even more relevant recommendations can be made.

[0007] Even though there are various ways available today for identifying the subset of listed items, e.g. recommendations, from a vast amount of available products, the technology used today sometimes nevertheless fail to present a relevant sub-set. This comes with the effect that the user is not buying or selecting any product, or that a product is bought or selected that is later returned or not used. Since today’s technology for finding recommendations often comes with the cost that the user has to accept to be traced, there is also a need for an approach that can provide recommendations to users that do not feel comfortable with having their activity being traced and stored.

[0008] It is an object of the invention to at least partly overcome one or more of the above-identified limitations of the prior art. In particular, it is an object to provide methods and apparatuses making it possible to provide a selection of listed items, pertaining to products and / or services, that matches needs and wishes of a user reflected in a user input obtained via a user interface of a client computer. More particularly, it is an object to provide such methods and apparatus at the same time as the user is provided with improved control of the selection process, thereby making it possible to select products and services that could fit not only the user’s own needs and wishes, but also with the needs and wishes of another person. Further, it is an object to provide a possibility to provide the selection of listed items, herein referred to as a sub-set, to the user without requiring tracing and storing of the digital footprint of the user.

[0009] Generally, it has been found that by using a machine learning approach, that is, using an artificial neural network or the like, and requesting user input, e.g. answers to questions and images reflecting interests of the user, and using the machine learning model for combining the user input with the listed items, representing products and services having attributes linked thereto, it is possible to provide the sub-set without the need for tracing the activity of the user. In addition, since the user does not have to be exposed to a large amount of listed items such that the activity can be traced, the process suggested herein for finding relevant items may be more time efficient for the user. In addition, the risk for the user of being deceived to select a product or service that was not initially intended may also be reduced if using the approach suggested herein.

[0010] According to a first aspect, it is provided a method for determining a sub-set of listed items from a main set of listed items held on a central server communicatively connected to a client computer. The listed items pertain to products and / or services, each of them being associated with different attributes. The method may comprise obtaining by the central sever an initiation request from the client computer, in response to the initiation request, transmitting a user input request from the central server to the client computer, in response to the user input request, obtaining, at the central server, user input retrieved at least partly via the client computer, ranking, at the central server, the listed items of the main set by using a machine learning model based on the user input obtained and the different attributes associated with the products and / or services pertaining to the listed items, assigning, at the central server, a number of top ranked listed items as the sub-set of listed items, and transmitting the sub-set of listed items from the central server to the client computer.

[0011] In line with the advantages described above, by allowing the user to provide the user input and to use the machine learning model for identifying the sub-set, representing a selection of the listed items identified as matching the user’s needs and wishes, by using the user input, e.g. answers to questions and / or images chosen by the user, and the attributes linked to the listed items, it is made possible to have the sub-set identified in a time efficient manner, compared to having recommendations identified by tracing user activity, and in a way in which the user will have a better understanding on the information underlying the selection. The wording “machine learning model” should be construed broadly and encompass any model that can be used for identifying relevant items based on historical data.

[0012] The number of top ranked listed items may be a non-user set number and may be less than 50, more specifically the number may be 24 or less.

[0013] By having the number of top ranked listed items, that is, the number of listed items forming part of the sub-set, non-user set, it is made possible to train the machine learning model accordingly. For instance, if having the number set to 20 listed items, the machine learning model can be trained to provide within this set, listed items that may fit the user input in different ways. By way of example, by having this limitation, 20 listed items, and being provided with the user input comprising images showing persons wave surfing, the sub-set may comprise both items related to wetsuits and surfboards to make sure that both categories are covered within the same set even if wetsuits items would be among the top 25 items in a situation in which the limitation is set to 20 items.

[0014] The initiation request may be user triggered.

[0015] To make sure that the user is aware that he or she is providing information in the form of the user input that will be used as input to the machine learning model that will in turn identify the sub-set, the initiation request may be user triggered, that is, an active decision by the user may be required for initiating the process.

[0016] The method may further comprise obtaining, at the central sever, a list item selection from the client computer, wherein the list item selection pertain to one item of the sub-set of listed items, transmitting a post-sales information request linked to the list item selection from the central server to the client computer and / or a logistic hub computer, wherein the post-sales information request pertains to information regarding whether or not the product or service linked to the list item selection has been returned or not, obtaining post sales information linked to the list item selection from the client computer and / or the logistics hub computer, fine-tuning the machine learning model based on the user input giving rise to the sub-set of listed items, the product or service linked to the list item selection and the post-sales information linked to the list item selection.

[0017] By using the post-sales information, e.g. information related to whether or not the product has been returned or not, it is made possible to not only take short time effects into account when identifying the sub-set by using the machine learning model, but also long term effects. By way of example, the machine learning model may be allowing the post-sales information to be taken into account to identify the sub-set that comprises listed items that the user will not only buy, but also decide to keep, or refrain from returning. Since different users may return products to different degree, user identification data may be taken into account as well. Another option to address that different users have different return patterns is to pose questions related to how the products or service have been used to the users via the client computers as part of the post-sales information request. By having this piece of information and also the return information, the long terms effects can be estimated even more accurately by the machine learning model.

[0018] The post-sales information may further comprise information about whether or not the product or service has been used or not, and / or information about whether or not the product or service has been reported to suffer from quality issues. As described above, return information, information about how the product or service has been used may allow the machine learning model to estimate the long term effects more accurately and also more reliably. By also including information related to quality issues in the post-sales information, this can be even further improved.

[0019] The step of obtaining, at the central server, the user input retrieved at least partly via the client computer may further comprise a sub-step of obtaining user identification data via the client computer.

[0020] By having access to the user identification data, this may be used for retrieving the user input from third parties. For instance, by having this data, images, videos, etc may be downloaded or accessed from social media platforms via the user’s own account or accounts belonging to other users that the user for different reasons would use as the user input. One reason for that another user’s account could be feasible to use is if a gift is to be purchased for the person linked to this other user’s account.

[0021] The user input may comprise the user identification data and information linked to the user identification data shared via one or several social media platforms.

[0022] The user input may comprise one or several images.

[0023] The user input may comprise position data of the client computer.

[0024] By taking into account the position data, either by having the user entering the data via the user interface or having the information automatically fetched without direct involvement by the user, it is made possible to further improve the selection process, that is, identifying the sub-set such that this is perceived as more relevant by the user. For instance, by taking the position data into account, the listed items related to the products and / or services that are often purchased, and optionally not returned and used, in the same part of the world can form part of the factors taken into account for making the selection.

[0025] The sub-set of listed items may each be provided with a text item generated by the machine learning model based on the user input.

[0026] In addition to selecting the sub-set, the machine learning model, which may comprise of different sub-modules, can be arranged to generate one or several text items for the different items of the sub-set. The text item(s) may provide a short explanation way this or these item(s) has / have been identified as a good fit for the user.

[0027] Further, in the step of ranking the listed items of the main set by using the machine learning model, availability information, retrieved from the logistics hub computer, linked to the different listed items may be used as input to the machine learning model.

[0028] By taking the availability information into account in this manner, it is made possible to rank products in stock higher compared to stocks currently not in stock. The availability information may be taken into account as one of several factors with the result that products not in stock are not necessarily ruled out, but may still be considered relevant if considering relevant with respect to other factors. For instance, a bag, being similar or identical to bags depicted in images forming part of the user input, may rank among the top ranked listed items even though the product is expected to be in stock in 1-2 days.

[0029] According to a second aspect it is provided a method for obtaining a sub-set of listed items. The sub-set may be a selection made from a main set of listed items held on a central server communicatively connected to a client computer. The listed items may pertain to products and / or services, each of them being associated with different attributes. The method may comprise obtaining a user interaction, pertaining to an initiation request, via a user interface of the client computer, in response to obtaining the user interaction, transmitting the initiation request from the client computer to a central server, obtaining, at the client computer, a user input request from the central server, in response to obtaining the user input request, transmitting user input from the client computer to the central server, wherein the user input is obtained via the user interface of the client computer, and in response to the user input, obtaining, at the client computer, the sub-set of listed items from the central server.

[0030] The same features and advantages as presented above with respect to the first aspect also apply to this second aspect.

[0031] According to a third aspect it is provided a server for determining a sub-set of listed items from a main set of listed items. The server may be communicatively connected to a client computer. The listed items may pertain to products and / or services, each of them being associated with different attributes. The server may comprise control circuitry configured to execute an initiation request obtaining function configured to obtain an initiation request from the client computer, a user input request transmitting function configured to transmit a user input request from the central server to the client computer in response to the initiation request, a user input obtaining function configured to obtain a user input retrieved at least partly via the client computer in response to the user input request, a ranking function configured to rank the listed items of the main set by using a machine learning model based on the user input obtained and the different attributes associated with the products and / or services pertaining to the listed items, an assigning function configured to assign a number of top ranked listed items as the sub-set of listed items, and a transmitting function configured to transmit the sub-set of listed items from the central server to the client computer.

[0032] The same features and advantages as presented above with respect to the first aspect also apply to this aspect.

[0033] According to a fourth aspect it is provided a client computer for obtaining a subset of listed items, wherein the sub-set may be a selection made from a main set of listed items held on a central server communicatively connected to the client computer, wherein the listed items pertain to products and / or services, each of them being associated with different attributes. The client computer may comprise control circuitry configured to execute a user interaction obtaining function configured to obtain a user interaction pertaining to an initiation request via a user interface of the client computer, an initiation request transmitting function configured to transmit the initiation request from the client computer to the central server, a user input request obtaining function configured to obtain a user input request from the central server, a user interface arranged for user interaction, a user input transmitting function configured to transmit a user input from the client computer to the central server, wherein the user input is obtained via the user interface of the client computer, and a sub-set of listed items obtaining function configured to obtain the sub-set of listed items from the central server.

[0034] The same features and advantages as presented above with respect to the second aspect also apply to this aspect.

[0035] According to a fifth aspect it is provided a computer program product comprising instructions which, when the program is executed by a computing device, causes the computer to carry out the method according to the first aspect.

[0036] According to a sixth aspect it is provided a computer program product comprising instructions which, when the program is executed by a computing device, causes the computer to carry out the method according to the second aspect.

[0037] Still other objectives, features, aspects and advantages of the invention will appear from the following detailed description as well as from the drawings. Brief Description of the Drawings

[0038] Embodiments of the invention will now be described, by way of example, with reference to the accompanying schematic drawings, in which

[0039] Fig. 1 generally illustrates communication between a client computer and a central server.

[0040] Fig. 2 is a flowchart illustrating a method for obtaining a sub-set of listed items performed in the client computer.

[0041] Fig. 3 is a flowchart illustrating a method for determining the sub-set of listed items performed in the central server.

[0042] Fig. 4 is a schematic illustration of the central server.

[0043] Fig. 5 is a schematic illustration of the client computer.

[0044] Detailed Description

[0045] Fig. 1 generally illustrates by way of example a system 100 comprising a client computer 102 and a central server 104. The client computer 102 can be any type of device that is equipped with a user interface and a configured for communication with the central server 104. The central server 104 may be any data processing apparatus that can communicate with other devices, such as the client computer 102, and also hold information about listed items pertaining to products and / or services. By way of examples, the central server 104 may be a server arranged to host a digital market place or similar. The client computer 102 can be a laptop, a mobile phone, a tablet or any other type of device capable of communicating with the central server and having a user interface such that all or a sub-set of the listed items held on the central server 104 can be displayed to a user of the client computer. As illustrated, the system 100 may also comprise a social media server 106 and a logistics hub server 108. The social media server 106 may be arranged to hold information in the form of text, images, videos etc linked to different users. The logistics hub server 108 may be arranged to hold information about products and / or services sold in the past, and also products and / or services returned in the past. Further, the logistics hub server may also comprise information about complaints made with respect to different products and / or services made in the past.

[0046] As illustrated, a process for determining a subset of listed items held on the central server 104 may start with that an initiation request 110 is transmitted from the client computer 102 to the central server 104. Once this is obtained, sometimes referred to as received, by the central server 104, a user input request 112 may be transmitted in the other direction, that is, from the central server to the client computer. The initiation request 110 may be a simple request that serves the sole purpose of starting the process, but it may also include additional information. For instance, in case the user has received a gift card, the initiation request may include a gift card number, which in turn may be associated with a specific sum of money. The gift card, or the initiation request, may also be linked to certain types of products. For instance, the gift card may not allow the gift card holder to buy products that are not deemed suitable for persons below a certain age.

[0047] In response to the user input request 112, user input 114 may be transmitted from the client computer 102 to the central server 104. The user input 114 may take various forms. For instance, the user input may be answers to questions provided via the user input request 112. In addition, the user input 114 may comprise images depicting activities the user enjoys doing, e.g. skiing, or products the user likes, e.g. a designer lamp the user finds beautiful. In addition, the user input 114 may also comprise user identification data, provided via the client computer, such that e.g. the images can be fetched from the social media server 106. The user input 114 may comprise the user’s own user identification data, e.g. his or her user name on one or more social media platforms, but it may also comprise other users’ user identification data on different social media platforms. For instance, if another user is often posting images that the user finds relevant for identifying products and / or services for the user, this other user’s user identification data may form part of the user identification data. In case the other user’s social media platform account is public, this may be done without a consent from the other user, but in case it is a non-public account, or in any other way restricted to use without permission, a consent may be requested and received before this other user’s social media information is made part of the user input.

[0048] Based on the user input 114, a sub-set of listed items 116 is selected at the central server 104 and transmitted to the client computer 102. By way of example, this sub-set 116 may comprise 20 items, that is, 20 products and / or services that has been found relevant by a machine learning model, such as an artificial neural network, with the user input 114 as part of the input.

[0049] Thus, unlike systems known today, all listed items on the central server 104 are not made available to the client computer 102. Instead, according to the approach presented herein, the user input 114 can be requested before any items have been presented to the user via the client computer 102. This is advantageous in that the user is more in control of the buying process and the risk of being distracted by other products or services can be reduced. Further, since the systems available today often builds on that all items are presented from the start and that the user’s activity is traced, such that the sub-set of listed items can be determined based on the user activity, the user input in these systems often develops over time. Even though there are advantages with this user activity approach, it comes with the disadvantage that users that are buying products for others, e.g. gifts, may not be provided with relevant recommendations. Further, since the recommendations made via systems tracing the users activity need to store data about the users over time, this may be an issue for some users. Put differently, since the user is not in control of the input used for providing the recommendations, situations may arise when the recommendations are not relevant, e.g. when the user is to buy a gift for someone else. In addition, since many of the systems used today are capturing information about the user’s preferences over time by tracing the activity, data must be collected over a period of time before relevant recommendations can be provided. This is a drawback with the systems of today.

[0050] As illustrated, after being provided with the sub-set of listed items 116, a list item selection 118, that is a selection related to one of the items of the subset, may be transmitted from the client computer 102 to the central server 104. The list item selection 118 may be based on input retrieved via the user interface of the client computer 102.

[0051] In response to the list item selection 118, the central server 104 may initiate an order such that the product or service linked to the list item selection 118 is provided to the user of the client computer or the person to whom the user ordered this product or service. In addition to initiating such order, a post-sales information request 120 may be transmitted from the central server 104 to the client computer 102. As illustrated, such post-sales information request 120 may also be transmitted to the logistics hub server 108. In response to this request, post-sales information 122 may be provided from the client computer 102 and / or the logistics hub server 108. The post-sales information 122 may comprise information about whether or not the product or service has been used or not, and / or information about whether or not the product or service has been reported to suffer from quality issues. The post-sales information 122, may in combination with the user input 114 give rise to the sub-set of listed items 116 and the product or service linked to the list item selection 118, be used for fine-tuning the machine learning model. By doing so, the sub-set of listed items are not only selected based on what will result in a sales, but also taking into account whether or not the product or service is returned or not. In addition, by also requesting information about whether or not the product or service has been used or not, also information about the products and / or services that are not returned, but still not used, can be taken into account, thereby reducing a risk of a sales of a product or service that is not being used. The post-sales information may be requested at more than one time, and it may also be requested weeks or months after the purchase.

[0052] In addition, an availability request may also be transmitted from the client computer 102 to the logistics hub server 108 either directly or indirectly via the central server 104, and availability information may be provided to the client computer and / or the central server 104 in response to this request. By way of example, the availability request may be a request for products in stock and the availability information may comprise this information. If being transmitted directly to the client computer, the availability information may be provided together with the sub-set of listed items. On the other hand, if being transmitted to the central server 104, this may be provided as one of several factors to take into account in the step of ranking the listed items. Put differently, the availability information may be provided as an add-on or it may be used as one of the factors being taken into account when ranking the listed items.

[0053] Even though it is illustrated only one logistics hub server, multiple such servers may be contacted with respect to the availability information. For instance, each retailer forming part of this ecosystem may have a logistics hub server on its own.

[0054] Having access to different retailers’ logistics hub servers comes with the additional benefit that if the user decides to order one of the products and / or services forming part of the sub-set of listed items, this order may be directly fed to the logistics hub server linked to the ordered product and / or service. In this way, the product and / or service may be delivered directly from the company offering the ordered product and / or service without any middlemen. Apart for the possibility to offer drop-shipping, by having these different servers closely linked in this way provides for that the risk that the availability information is incorrect, or more correctly obsolete, is lowered.

[0055] Further, the services pertaining to the listed items may comprise gift cards or any other type of voucher connected to one or several parties. Hence, the different attributes combined with the user input during the ranking do not necessarily link to a specific product or a specific service, but can be attributes associated with a company providing several products and / or services linked to the user input. For instance, in case the user input suggest that there is an interest in skiing and mountaineering, the products and / or services may include a gift card to a retailer offering goggles, boots and other equipment related to skiing and mountaineering, but also gift cards to travel agencies offering trips to the Alps.

[0056] Fig. 2 is a flowchart illustrating a method 200 performed in the client computer 102 and an interrelated method 300 performed in the central server 104 by way of example are illustrated in fig. 3. As illustrated, as a first step, a user interaction may be obtained 202 at the client computer 102. This user interaction may be obtained via a user interface of the client computer. The user interaction may have a variety of forms. By way of example, the user interaction may be that the user is clicking a link or a button on a website such that a process of selecting the sub-set of listed items are started. The user interaction may be a stand-alone interaction or it may form part of another interaction as well. For instance, in case the user is to select a gift for herself or for someone else, the user interaction may form part of a step in which the user is entering her gift card number on a website.

[0057] Once the user interaction is obtained, the initiation request 110 is transmitted 204 from the client computer 102 to the central server 104. In line with the example above, this request may, in addition to a consent from the user to start the process, comprise the gift card number or other similar data.

[0058] At the central server 104, the initiation request 110 can be obtained 302 and in response to receiving this, the user input request 112 may be transmitted 304 back to the client computer. The user input request 112 may be questions for the user to answer that will serve the purpose of providing a selection of listed items that fit the needs or wishes of the user. As described above, the user input may include images, social media platform references, etc, and the type of information requested from the user may be reflected in the user input request 304. For instance, in case a company providing the gift card has a policy to not use certain social media platforms or the like, such restrictions may be reflected in the initiation request and also in the user input request.

[0059] At the client computer 102, the user input request may be obtained 206, and based on this request, the user input may be obtained via the user interface. After having obtained the user input, the user input may be transmitted 208 to the central server 104. As described above, the user input may take different forms. It may be text input, e.g. answers to questions provided via the user input request, or it may be images, that also may form part of answers to the questions forming part of the user input request. Further, the user input 114 may also include references to material held by third parties. For instance, as described above, the user input may include user identification data such that information related to a specific user on a social media platform can be retrieved.

[0060] The user input may be obtained 306 at the central server. As discussed, as a sub-step of the step of obtaining the user input, the user identification data can be obtained.

[0061] Based on the user input, listed items of a main set held in the central server can be ranked 308. Put differently, by having the user input provided by the user via the client computer, items that are found likely to be selected or purchased by the user is identified. By having attributes linked to the different products and / or services represented by the items, a multitude of factors can be taken into account. The attributes may not only be restricted to category, colour, material, etc, but may also include e.g. images or videos used in marketing material for the different products and / or services. The ranking may be performed by the machine learning model as described above. The machine learning model may however not be restricted to the specific step of ranking, but can also be used for generating the user input request. Further, even if illustrated that the user input request is transmitted once from the central server 104 to the client computer 102, additional user input requests may be made if additional user input may found beneficial to provide a relevant sub-set of listed items. Whether or not additional user input requests are to be made and also what information to be requested in such additional requests may be determined by the machine learning model.

[0062] Once having the listed items ranked, a number of top ranked listed items are assigned 310 as the sub-set of listed items. Thereafter, the sub-set of listed items can be transmitted 312 to the client computer 102. At the client computer 102, the sub-set of listed items can be obtained 210.

[0063] After having had the sub-set of listed items displayed or in any other way made available to the user, a list item selection pertaining to one, or more, of the sub-set of listed items can be obtained via the user interface and transmitted 212 to the central server 104. At the central server 104, the list item selection can be obtained 314. Once being obtained, an order may be initiated such that the product and / or service is delivered. Further, the post-sales information request may be transmitted 316 to the client computer 102 and / or the logistics hub server 108. In response to this request, the post-sales information may be obtained 318. As described above, the post-sales information may be used for training, or fine-tuning 320, the machine learning model such that more relevant sub-set selections can be made going forward. Further, the fine-tuning of the machine learning model may also result in that more relevant user input requests are made, in turn providing for more relevant user input. In addition, the machine learning model may also be fine-tuned such that the post-sales information requests are made more relevant. For instance, the fine-tuning of the machine learning model may result in that that a time period between the time of obtaining the list item selection and a time of transmitting the post-sales information request is changed, and also that the other type of information is requested in the post-sales information request.

[0064] Fig. 4 is a schematic illustration of the client computer 102 and fig. 5 is a schematic illustration of the central server 104 communicatively connected to the client computer 102.

[0065] As described above the client computer 102 may be configured for obtaining the sub-set 116 of listed items, wherein the sub-set is a selection made from the main set of listed items held on the central server 104. The listed items may pertain to products and / or services, each of them being associated with different attributes. The client computer 102 may comprise control circuitry 500 configured to execute a user interaction obtaining function 502 configured to obtain the user interaction pertaining to the initiation request 110 via the user interface 512 of the client computer 102, an initiation request transmitting function 504 configured to transmit the initiation request 110 from the client computer 102 to the central server 104, a user input request obtaining function 506 configured to obtain a user input request 112 from the central server 104, the user interface 512 arranged for user interaction, a user input transmitting function 508 configured to transmit a user input 114 from the client computer 102 to the central server 104, wherein the user input 114 may be obtained via the user interface 512 of the client computer 102, and a sub-set of listed items obtaining function 510 configured to obtain the sub-set 116 of listed items from the central server 104, wherein the sub-set of listed items is determined at the central server 104 by using a machine learning model 414 provided with the user input 114 and the different attributes of the products and / or services pertaining to the listed items.

[0066] The server 104 for determining the sub-set 116 of listed items from the main set of listed items is illustrated in fig. 4. The server 104 may comprise control circuitry 400 configured to execute an initiation request obtaining function 402 configured to obtain an initiation request 110 from the client computer 102, a user input request transmitting function 404 configured to transmit a user input request 112 from the central server 104 to the client computer 102 in response to the initiation request 110, a user input obtaining function 406 configured to obtain the user input 114 retrieved at least partly via the client computer 102 in response to the user input request 112, a ranking function 408 configured to rank the listed items of the main set by using the machine learning model 414 based on the user input 114 obtained and the different attributes associated with the products and / or services pertaining to the listed items, an assigning function 410 configured to assign a number of top ranked listed items as the sub-set of listed items 116, and a transmitting function 412 configured to transmit the sub-set 116 of listed items from the central server 104 to the client computer 102.

[0067] From the description above follows that, although various embodiments of the invention have been described and shown, the invention is not restricted thereto, but may also be embodied in other ways within the scope of the subject-matter defined in the following claims.

Claims

CLAIMS1 . A method (300) for determining a sub-set of listed items (116) from a main set of listed items held on a central server (104) communicatively connected to a client computer (102), wherein the listed items pertain to products and / or services, each of them being associated with different attributes, said method comprising obtaining (302) by the central sever (104) an initiation request (110) from the client computer (102), in response to the initiation request (110), transmitting (304) a user input request (112) from the central server (104) to the client computer (102), in response to the user input request (112), obtaining (306), at the central server (104), user input (114) retrieved at least partly via the client computer (102), ranking (306), at the central server (104), the listed items of the main set by using a machine learning model (414) based on the user input (114) obtained and the different attributes associated with the products and / or services pertaining to the listed items, assigning (308), at the central server (104), a number of top ranked listed items as the sub-set of listed items (116), and transmitting (310) the sub-set of listed items (116) from the central server (104) to the client computer (102).

2. The method according to claim 1 , wherein the number of top ranked listed items is a non-user set number and less than 50.

3. The method according to any one of the preceding claims, wherein the initiation request (110) is user triggered.

4. The method according to any one of the preceding claims, further comprising obtaining (314), at the central sever (104), a list item selection (118) from the client computer (102), wherein the list item selection (118) pertain to one item of the sub-set of listed items (116), transmitting (316) a post-sales information request (120) linked to the list item selection (118) from the central server (104) to the client computer (102) and / or alogistic hub computer (108), wherein the post-sales information request (120) pertains to information regarding whether or not the product or service linked to the list item selection (118) has been returned or not, obtaining (318) post sales information (122) linked to the list item selection (118) from the client computer (102) and / or the logistics hub computer (108), fine-tuning (320) the machine learning model (414) based on the user input (114) giving rise to the sub-set of listed items (116), the product or service linked to the list item selection (118) and the post-sales information (122) linked to the list item selection (118).

5. The method according to claim 4, wherein the post-sales information (122) further comprises information about whether or not the product or service has been used or not, and / or information about whether or not the product or service has been reported to suffer from quality issues.

6. The method according to any one of the preceding claims, wherein the step of obtaining (306), at the central server (104), the user input (114) retrieved at least partly via the client computer (102) further comprises a sub-step of obtaining (307) user identification data via the client computer (102).

7. The method according to any one of the preceding claims, wherein the user input (114) comprises the user identification data and information linked to the user identification data shared via one or several social media platforms.

8. The method according to any one of the preceding claims, wherein the user input (114) comprises one or several images.

9. The method according to any one of the preceding claims, wherein the user input (114) comprises position data of the client computer (102).

10. The method according to any one of the preceding claims, wherein the subset of listed items (116) are each provided with a text item generated by the machine learning model (414) based on the user input (114).

11. The method according to any one of the preceding claims, wherein, in the step of ranking (306) the listed items of the main set by using the machine learning model (414), availability information, retrieved from the logistics hub computer (108), linked to the different listed items is used as input to the machine learning model.

12. A method (200) for obtaining a sub-set of listed items (116), wherein the sub-set is a selection made from a main set of listed items held on a central server (104) communicatively connected to a client computer (102), wherein the listed items pertain to products and / or services, each of them being associated with different attributes, said method comprising obtaining (202) a user interaction, pertaining to an initiation request (110), via a user interface (512) of the client computer (102), in response to obtaining the user interaction, transmitting (204) the initiation request (110) from the client computer (102) to a central server (104), obtaining (206), at the client computer (102), a user input request (112) from the central server (104), in response to obtaining the user input request (112), transmitting (208) user input (114) from the client computer (102) to the central server (104), wherein the user input (114) is obtained via the user interface (512) of the client computer (102), and in response to the user input (114), obtaining (210), at the client computer (102), the sub-set of listed items (116) from the central server (104).

13. A server (104) for determining a sub-set (116) of listed items from a main set of listed items, wherein the server (104) is communicatively connected to a client computer (102), wherein the listed items pertain to products and / or services, each of them being associated with different attributes, said server (104) comprising control circuitry (400) configured to execute an initiation request obtaining function (402) configured to obtain an initiation request (110) from the client computer (102), a user input request transmitting function (404) configured to transmit a user input request (112) from the central server (104) to the client computer (102) in response to the initiation request (110), a user input obtaining function (406) configured to obtain a user input (114) retrieved at least partly via the client computer (102) in response to the user input request (112),a ranking function (408) configured to rank the listed items of the main set by using a machine learning model (414) based on the user input (114) obtained and the different attributes associated with the products and / or services pertaining to the listed items, an assigning function (410) configured to assign a number of top ranked listed items as the sub-set of listed items (116), and a transmitting function (412) configured to transmit the sub-set (116) of listed items from the central server (104) to the client computer (102).

14. A client computer (102) for obtaining a sub-set (116) of listed items, wherein the sub-set is a selection made from a main set of listed items held on a central server (104) communicatively connected to the client computer (102), wherein the listed items pertain to products and / or services, each of them being associated with different attributes, said client computer (102) comprising control circuitry (500) configured to execute a user interaction obtaining function (502) configured to obtain a user interaction pertaining to an initiation request (110) via a user interface (512) of the client computer (102), an initiation request transmitting function (504) configured to transmit the initiation request (110) from the client computer (102) to the central server (104), a user input request obtaining function (506) configured to obtain a user input request (112) from the central server (104), a user interface (512) arranged for user interaction, a user input transmitting function (508) configured to transmit a user input (114) from the client computer (102) to the central server (104), wherein the user input (114) is obtained via the user interface (512) of the client computer (102), and a sub-set of listed items obtaining function (510) configured to obtain the subset (116) of listed items from the central server (104).

15. A computer program product comprising instructions which, when the program is executed by a computing device, causes the computer to carry out the method (300) according to any one of the claims 1 - 11.

16. A computer program product comprising instructions which, when the program is executed by a computing device, causes the computer to carry out the method (200) according to claim 12.

Images