JP2024041849A - Probabilistic item matching and search - Google Patents

Abstract

Systems, methods and computer program products for probabilistic item matching and retrieval are provided.
A system including at least one processor receives data points related to a particular item, generates a database query based on the data points, and receives a response to the database query. The response includes multiple candidate items related to the particular item. The system also receives a first input associated with a particular item, generates probability scores for at least two of the plurality of candidate items in the response based on the at least second input, and generates probability scores for at least two candidate items of the plurality of candidate items in the response, and Selects a selected item from the items based on the selected item's probability score and outputs a reference to the selected item or a value representing the selected item.
[Selection diagram] Figure 4

Description

Cross-reference of related applications
[0001] This application is filed in U.S. Patent Application No. 62/740,165, entitled “Probabilistic Search Biasing and Recommendations,” filed on October 2, 2018; U.S. Patent Application No. 62/740,182 entitled “Probabilistic Item Matching and Searching” filed February 28, 2019 “Probabilistic Search Biasing and Recommendation” U.S. Patent Application No. 16/288, entitled No. 373, and U.S. patent application Ser. No. 16/288,199 entitled “Inventory Ingestion, Image Processing, and Market Descriptor Pricing System,” filed February 28, 2019. Ingestion and Pricing System” U.S. Patent Application No. 16/288,203, entitled “Determining Sellability Score and Cancellability Score,” filed February 28, 2019, and U.S. Patent Application No. 16/288,158, filed February 28, 2019; Concerning U.S. patent application Ser. Incorporated herein It is.

[0002] This disclosure generally relates to computer-implemented search and recommendation generation based on probabilistic analysis of large data sets.

[0003] Often, when interested in buying or selling an item via an online platform, finding that item and items similar to it can pose a challenge. For example, describing items for sale in a way that does not result in multiple different listings for the same item can be a challenge for both vendors and vending platforms. Similarly, it can be confusing for buyers searching for a specific item to receive multiple different results for what could be the same item, and the reason for this is that the purchased item does not match the item description during the first transaction. This is because they will receive the following. Thus, buyers and sellers have struggled with the accuracy of search results and matching items.

[0004] A related issue concerns providing and receiving relevant search results within the terms of service or other meaningful constraints of a given trading platform. For example, all else being equal, a buyer may want to receive an item as soon as possible, and the platform will pair the buyer with a seller who can ship the item and have it delivered as soon as possible. You may want to provide quality of service (QoS) levels or service level agreements (SLAs). Similarly, sellers need to know how to price an item in order to move it within a certain time window, for a certain profit, or to gain the possibility of closing a sale. There may be no reliable way to guess how an item should be priced. In the C2C market, which includes consumer-to-consumer (C2C) e-commerce, both buyers and sellers want to make these aspects of marketing easier, making these easier solutions even more transparent and less labor intensive. We have seen a long-standing need for a smart platform.

[0005] The accompanying drawings, which are incorporated in and form a part of this specification, illustrate embodiments of the disclosure and, together with the description, explain the principles of the disclosure and will be understood by those skilled in the art. further assists in enabling the creation and use of the embodiments.

[0006] FIG. 1 illustrates an example system useful for implementing methods including probabilistic item matching and searching, according to some embodiments. [0007] FIG. 2 illustrates an example system useful for implementing methods including probabilistic search biasing and recommendations, according to some embodiments. [0008] FIG. 4 illustrates an example screen display including an example prompt to enter an amount, according to some embodiments. [0009] FIG. 2 is a flowchart illustrating a method including probabilistic item matching and searching, according to some embodiments. [0010] FIG. 2 is a flowchart illustrating a method including probabilistic search biasing and recommendations, according to some embodiments. [0011] FIG. 1 is a block diagram of a general purpose computer that can be used to implement various aspects of the present disclosure.

[0012] In the drawings, like reference numbers generally indicate the same or similar elements. Additionally, the left-most digit of a reference number generally identifies the drawing in which the reference number first appears.

[0013] Systems, apparatus, devices, methods, and/or computer program products for providing probabilistic item matching, searching, search biasing, recommendations, and/or any combination thereof are described herein. Embodiments and/or combinations and subcombinations thereof are provided.

[0014] FIG. 1 is a diagram illustrating an example system useful for implementing methods including probabilistic item matching and searching, according to some embodiments.

[0015] On the left side of FIG. 1 is a depiction of a merchant and sales application (app) that, in some embodiments, displays posted items for sale. and can be used to create new listings of items for sale. These depictions include the front-end design of the app, which can be a web app, mobile app (e.g., native, web-based, or hybrid), and can include various levels of user interaction. Interactions with buyers rather than sellers may, in some embodiments, include similar interactions with apps.

[0016] To the right of center of FIG. 1 is a depiction of an example backend architecture that may be useful for implementing probabilistic item matching and search, according to one embodiment. Collectively, the components illustrated in FIG. 1 can be treated as comprising, among other things, example system 100, which includes data (item data, metadata, listings, databases, etc.) in the back end. ), and the processing of inputs, outputs, results, and related interactions at the front end.

[0017] In addition to interactions with users (eg, buyers/sellers), the backend may process input from users and/or various other sources. For example, if a seller provides input regarding an item for a new listing, the backend may attempt to verify the identity of what the seller is listing via other inputs. Other inputs may be prompted by the merchant, automatically derived from other sources, or a combination of both in some embodiments.

[0018] To generate an offer price recommendation, the backend may (1) match older listings in the backend database of listings, such as on the same platform; and (2) match older listings on the same platform and third party. (3) other C2C platforms, business-to-consumer (B2C) outsourcing sites, and/or retail platforms; (4) any information obtained from (1), (2), (3), or any combination thereof; , verifying with the merchant, including prompting for input to confirm whether the merchant's information is correct, can be performed in any order.

[0019] Receiving and parsing user input regarding the item, whether or not, or checking any of (1), (2), (3), or any combination thereof, as described above; Any matching, whether or not, may be performed at any stage of such processing as described herein. More specifically, the term "matching" may refer to more than determining that text strings or other data sequences match.

[0020] For example, matching in any of the above scenarios may include the use of natural language processing (NLP) techniques, including the use of synonyms (e.g., synonym libraries or lookup tables, optional abbreviations), etc. further searches for matches to the found synonyms; identify any other semantically related words; and/or further searching and performing additional matches, performing word stemming based on a minimum of one language and a corresponding set of morphological rules, identifying misspellings and equating them with correct spellings ( or otherwise correcting misspellings) and the like for similar processing. Thus, in some embodiments, matching may be recursive. Similar linguistic analysis of input or item descriptions can be used to separate modifiers, e.g., adjectives that describe size or color, from other name-related identifiers, e.g., nouns that designate names or items. can.

[0021] Furthermore, matching can be performed not only on generic or categorical descriptions, but also on proper nouns, eg, brand names, commonly named trademarks, and the like. In some embodiments, a classification algorithm or other neural network or artificial intelligence is used to search the search query, user input, or item description obtained from (1), (2), or (3) above. It is possible to extract related words from an unstructured word set such as In this way, a given classifier is able to "understand" terms from a given unstructured word set that can be intended as brand names compared to generic descriptions, e.g. Any database, library, lookup table, etc. can be configured to update to include in a particular category or subcategory as well as in a particular category or subcategory to improve matching. The matching algorithm or any component may be biased or adjusted with respect to precision or recall.

[0022] In some embodiments, matching as described above is performed locally on a client device, e.g., a personal computer (PC) or a mobile device such as a smartphone or tablet computer, using a self-hosted service (a dedicated server, It can be run remotely on infrastructure (including a virtual private server (VPS), or on-premises cloud), remotely through a third-party service, or any combination thereof. In embodiments where matching operations can be performed across multiple devices, certain matching operations are performed locally and intermediate results are sent to a remote service for confirmation, validation, and/or other further processing. A combination of these can be done.

[0023] In one embodiment, one method of identifying the item or type of item may be to resolve the proposed new listing to a unique identifier, which may be, for example, This is done, for example, by comparison with items that have already been mapped to unique identifiers. Information about items identified in this way can then be more reliably correlated, for example, with information on other comparable items listed for sale in a back-end database, or with other sources. Search biasing and price recommendations can be provided accordingly, as further described below with respect to FIGS. 2 and 5, and incorporated herein by reference, cited above. 16/288,373, entitled "Probabilistic Search Biasing and Recommendations," published by J.D.

[0024] For highly unique items (e.g., one-of-a-kind items, custom-made items, craft items, personalized items, etc.) that may not have a corresponding unique identifier to other items of the same type. Some items may not have unique identifiers mapped to other similar items. In such cases, the backend may proceed differently (not shown), for example, in some example embodiments, duplicate listings may be sent to other platforms (e.g., for custom or craft items). ) or immediately prompt the seller to verify such uniqueness.

[0025] The reference or value may, in some embodiments, be a stock-keeping unit (SKU), a uniform product code (UPC), a uniform resource identifier (URI), a uniform resource locator to which a given item may be mapped. Child (URL) be a uniform, universal, and/or unique identifier, including but not limited to a Uniform Resource Name (URN), International Standard Book Number (ISBN), Amazon Standard Product Number (ASIN), etc. be able to. Additionally or alternatively, the value may be a first input or a second input (e.g., a string item description, checksums, fingerprints, signatures, digests, hashes, cryptographic hashes, etc., corresponding to a specific text field for brand name, size, enumerated selector such as model year, etc. For purposes of FIG. 1 and the accompanying discussion, the mapping of a given item to a unique identifier may also be referred to as SKU-level matching or SKU-level data.

[0026] At the initial input stage, or at any subsequent input stage, such as action (4), the system 100 may solicit additional user input, eg, via an app. The input can be, for example, text in the form of a string of characters, a photographic image, voice recognition, other characteristic sounds or audio fingerprints. Text-based input through the app may prompt users to fill out a small number of information fields depending on the broad category of item they are listing for sale, rather than an exhaustive description of all features. In many cases, even basic textual information allows the backend to unravel SKU-level data, or at least sell a small number of possible candidate items for confirmation of correct item identification. It may be possible to present the information to a person.

[0027] In addition to text-based strings or images, mathematical or cryptographic signatures in any representation may also be used. Bar codes or other patterns or sequences can be used. Other possible types of input include, but are not limited to, vibration patterns, chemical samples and their analysis, measurements of radiation, electrical and/or magnetic signals, or any other environmental sensor input.

[0028] Other forms of input may be received, such as output from other computer programs or algorithms, such as in the form of neural network output, perceptron output, image recognition output, classification output, or other types of output. . These outputs may be based on other user inputs, such as photo data, voice or audio data, or other third party resources, feeds, etc. Any or all such input may be generated by a tablet computer or a mobile device such as a smartphone in some embodiments.

[0029] In the case of image input, a user attempting to create a new listing for sale may be prompted to take at least one photo of the item being listed. The photo image data for that photo can be processed locally on the device running the app or sent to a backend or third party for initial and/or further processing. Processing photographic image data can include the use of artificial intelligence. Processing the image data can include, for example, feeding the image data to a neural network, perceptron, or classifier. In some embodiments, image data may be processed using computer-implemented image recognition, such as using at least one computer vision algorithm, for example. Any of the techniques described above or their equivalents may be used to perform operations such as, for example, classification, object recognition, and/or reverse image retrieval, to name a few non-limiting examples. .

[0030] Any neural network described herein can include at least one artificial neural network (ANN). The ANN may include at least one of a forward propagation neural network, a recurrent neural network, a modular neural network, or a memory network, to name a few non-limiting examples. For example, for a forward-propagation neural network, this may in some embodiments be at least one of a convolutional neural network, a stochastic neural network, a time-delay neural network, a multilayer perceptron, an autoencoder, or any combination thereof. It can further correspond to one. Such ANNs may have multiple layers, and in some embodiments these layers may be tightly coupled, which may result in, for example, This may be the case if the activation function can be reused, or if the activation of one or more layers is skipped to solve a particular computation.

[0031] In further embodiments based, for example, on convolutional neural network (CNN) processing, such CNN may further integrate at least some filters (e.g., edge filters, horizon filters, color filters, etc.) . These filters may include, for example, some of edge detection algorithms, color filtering algorithms, and/or predetermined thresholds. For a desired result, a given ANN, including a CNN, may be designed, customized, and/or modified in various ways according to a number of criteria, in some embodiments, for example.

[0032] Image recognition herein may also utilize machine learning in some embodiments. Image recognition systems can be trained, for example, using either exhibits from a backend database, data partner feed data, exhibits scraped from public sources, or any other source of accurate training data. I can do it. Image recognition can be further configured to detect and interpret barcodes, quick response (QR) codes, labels, tags, logos, trademarks, and/or any other defining characteristics of the item. In some embodiments, the image recognition may further perform optical character recognition (OCR) and interpret the text using natural language processing (NLP). In this way, image recognition can be used to identify items to be put up for sale, or at least candidate items, that can be reviewed and selected based on new information and calculations that can later be introduced via various input sources. You can update the selected items.

[0033] As discussed above, actions (1)-(4) taken by system 100 may be performed in any order, with any performed or skipped based on any other automatic or manual determinations. It can be executed with However, considering reliability, computational efficiency, time, and possibly cost (e.g., accessing a particular platform), one default configuration is to initially (1) (1) is sufficient (e.g., provides a data point for a list of candidate items and/or other data points to confirm or update the selected item) , actions (2)-(4) can be bypassed and skipped to any update or confirmation actions. Additionally, subsequent actions can be "smart" or take into account responses or results from previous actions as they become available.

[0034] Similarly, if (1) does not satisfy the condition, or if further input is separately required, the system 100 proceeds to perform action (2) or (3), depending on which is more efficient. You may proceed depending on your current situation or other factors. Additionally or alternatively, in some embodiments, actions (1)-(4) may all be performed in parallel for potentially faster resolution.

[0035] Depending on the given platform and/or given merchant, merchant input may be treated as less reliable than input collected from any of (1)-(3). I can do it. In some embodiments, if any or all of (1)-(3) are still insufficient for any reason or for some other reason, optionally proceed with (4), e.g. may request Seller input, confirmation, or adjustment. The action taken in (4), in some embodiments, includes further details describing the item to be listed and/or confirmation whether the selected item or other candidate items may match the item to be listed. and providing further prompts for input from the merchant.

[0036] Once the system 100 "understands" what an item is, such as through artificial intelligence, image recognition, or other means available to the system 100, it may perform additional calculations to calculate other data or Items may be analyzed and categorized based on metadata, such as price, condition, marketability trends, etc. Further details regarding these additional calculations are discussed below with respect to FIGS. 2 and 5 and are incorporated by reference in the application entitled "Probabilistic Search Biasing and Recommendations" (U.S. Application No. 16/288). , 373).

[0037] FIG. 2 is a diagram illustrating an example system useful for implementing methods including probabilistic search biasing and recommendations, according to some embodiments.

[0038] On the left side of FIG. 2 is a depiction of a merchant and sales application (app) that, in some embodiments, is used to display listed items for sale and Can be used to create new listings for items. These depictions include the front-end design of the app, which can be a web app, mobile app (e.g., native, web-based, or hybrid), and can include various levels of user interaction. Interactions with buyers rather than sellers may, in some embodiments, include similar interactions with apps.

[0039] To the right of center of FIG. 2 is a depiction of an example backend architecture that may be useful for implementing probabilistic search biasing and recommendations, according to one embodiment. Collectively, the components illustrated in FIG. 2 can be treated as comprising, among other things, example system 200, which includes data (item data, metadata, listings, databases, etc.) in the back end. ), and the processing of inputs, outputs, results, and related interactions at the front end.

[0040] Besides interactions with users (e.g., buyers/sellers), the backend can process inputs from the user and/or various other sources. For example, if a seller provides input regarding an item for a new listing, the backend can seek to ascertain, via other inputs, an identification of what the seller is listing and its typical or comparable listing price, e.g., on the same market platform, other market platforms, consignment shops, retail stores, or combinations thereof. The other inputs, in some embodiments, can be prompted by the seller, automatically derived from other sources, or a combination of both.

[0041] In one embodiment, one method of collecting prices may be to start with items that have already been confirmed to match a given item in the database. An example of this is the application entitled "Probabilistic Item Matching and Searching" (U.S. Application No. 16/288,379), which is described herein with respect to FIGS. 1 and 4 and is incorporated by reference. further described. For items so verified, any corresponding price information can be expected to correlate more reliably with the item the seller is offering for sale. Price recommendations can be generated accordingly.

[0042] Conversely, unique items that may not have a corresponding unique identifier to other items of the same type (e.g., one-of-a-kind items, custom-made items, crafted items, personalized items, etc.) For example, the backend can proceed differently (not shown); for example, in some exemplary embodiments, it may proceed differently within the same broad category for unique items (e.g., based on broad category user input). Refuse to introduce your listings to sellers, automatically generate prices for such unique items based on broader categories or comparable platforms, and/or provide comparable or suggested price recommendations.

[0043] To generate an offer price recommendation, the backend, in some embodiments, (1) matches old listings in the backend database of listings; and (2) matches the same platform and third-party listings. (3) including matching data partner feed data, such as through subscription services with party providers, and (3) other C2C platforms, business-to-consumer (B2C) outsourcing sites, and/or retail platforms. (4) back-end information for verifying prices or as to why and how much the seller may wish to readjust prices; Confirm with the Seller, including prompting for input to re-adjust prices, possibly receiving automatically generated advice from I can do it.

[0044] In any of steps (1), (2), or (3) of system 200, price information may be collected and statistically analyzed with respect to other parameters or variables, and may include other parameters or variables. may include, for example, any other data or metadata about price, correlated with listing condition (new, like new, very good, good, fair, poor, etc.), category, brand, size, etc. Any differences in metadata that may occur on other platforms, whether on the same platform or on other platforms such as third-party or public sources (item condition grade , regional variations in pricing, seasonal variations in pricing, etc.). Statistical analysis here can include, for example, averaging or calculating trends for any set of these data points or metadata categories.

[0045] As discussed above, actions (1)-(4) taken by system 200 may be performed in any order, with any performed or skipped based on any other automatic or manual determinations. It can be executed with However, considering reliability, computational efficiency, time, and possibly cost (e.g., accessing a particular platform), one default configuration is to initially (1) (1) is sufficient (e.g., provides a data point for a list of candidate items and/or other data points to confirm or update the selected item) , actions (2)-(4) can be bypassed and skipped to any update or confirmation actions.

[0046] Similarly, if (1) does not satisfy the condition, or if further input is otherwise required, system 200 may proceed to perform action (2) or (3), depending on which is more efficient. You may proceed depending on your current situation or other factors. Additionally or alternatively, in some embodiments, actions (1)-(4) may all be performed in parallel for potentially faster resolution.

[0047] Depending on the given platform and/or given merchant, merchant input may be treated as less reliable than input collected from any of (1)-(3). I can do it. In some embodiments, if any or all of (1)-(3) are still insufficient for any reason or for some other reason, optionally proceed with (4), e.g. may request Seller input, confirmation, or adjustment. The action taken in (4) may, in some embodiments, further include providing automatically generated advice from the backend as to why and by what amount the seller may wish to readjust the price. can.

[0048] With respect to automatic biasing of search results, the same metadata that is considered regarding price can be used in determining where new or existing listings for sale should be displayed in a given set of search results. can be considered. price, probability score (e.g., regarding the likelihood that an item will sell within a given time period), item condition, geography, season, time of day, cyclical or long-term trends, larger-scale trends, and/or ( Upgrade (promote, increase, or may be otherwise increased) or downgraded (demoted, lowered, or otherwise decreased). Other factors in biasing search results may depend on the individual purchaser searching or the application programming interface (API) or medium through which the search can be conducted or results can be obtained.

[0049] Some use cases for biasing may be to improve shipping and/or delivery times for sold items by biasing near results over long distance results. good. Here, "short distance" refers not only to geographical distance, but also, for example, to "delivery distance", i.e. the average transit time of a parcel along a particular route or within/between a particular geographical area. It may refer to Other factors include the similarity of different items, which may be from nearby sellers, how long a given item has been listed for sale, and whether a given buyer or other buyers This may include whether they have expressed interest in the item (e.g., via a given platform), and so on. In some cases, sellers may also be given some control over search biasing, for example, in some embodiments, during certain time periods or based on seller reputation, conduct (karma), goodwill, etc. Guaranteed promotions and more.

[0050] For example, for a buyer searching for a specific item that may match a unique identifier to multiple instances around the world, search engines may, in some cases, Proximity sellers (with delivery distance closer to the buyer) may be surfaced over long-distance sellers even though they may have other attributes that may be of interest. In some embodiments, search biasing can be configured, for example, to override or weight more heavily than existing user-specified search filters or settings.

[0051] Other factors that search engines may consider include shipping costs for sellers and/or platforms, sales taxes for buyers and/or platforms, and costs for sellers and/or platforms, among other possibilities. Profitability can be included. Although such search biasing techniques can be beneficial to C2C platforms, in some use cases they can provide similar benefits to B2C or B2B platforms.

[0052] Such preferential search biasing may be further useful to market platforms, in further examples, to maintain QoS or adhere to SLAs. Thus, using at least the enhanced techniques described herein, some embodiments can improve searchability with respect to certain attributes of buyers and/or sellers with respect to items being traded. I can do it.

[0053] FIG. 3 illustrates an example of an exemplary user interface screen display 300 that may be rendered on a computing device that includes a display device, according to some embodiments. Screen display 300 may include various areas for outputting textual values and/or graphical elements, as well as prompts for input in various formats. For example, user input may be provided via any graphical user interface (GUI) element, text user interface (TUI), voice command, accelerometer, other environmental sensor, or the like.

[0054] The screen display provided in FIG. 3 is illustrative only and may be associated with some of the probabilistic item matching, search, search biasing, and recommendation enhanced techniques described herein. Example output and input are provided to illustrate. Those skilled in the relevant art will appreciate that various approaches can be taken to provide a suitable screen display 300 in accordance with this disclosure.

[0055] Screen display 300 may include various indicators and control elements that may be displayed and arranged according to any suitable methodology, framework, toolkit, widget set, guideline, design language, etc. Examples include flat design, hierarchical design, layered design, WIMP design (windows, icons, menus, pointers), natural user interface (NUI), reality-based interface (RBI), head-up display (HUD), etc. This includes augmented reality (AR), virtual reality (VR) using virtual storefronts, etc. User input can be via pointing devices, such as mice, trackballs, joysticks, touch pads, touch screens, etc., as well as other forms of touch, spatial navigation, eye tracking, voice commands, environmental sensors, scripts or programs via APIs. It can be provided using input etc. Equivalent input means understood in the art may additionally or alternatively be used.

[0056] Screen display 300 may include a title, such as that displayed at the top of screen display 300 in FIG. 3, in some embodiments. Other areas of screen display 300 may include, for example, other GUI elements, widgets, knobs, sliders, scroll bars, buttons, or related functionality for the user to interact with any underlying hardware and/or software. can be presented.

[0057] For further explanation, an example is shown in FIG. In the example shown in FIG. 3, $990 is a "proposed price" that represents an example of the proposed amount output on the screen display. Additionally, a prompt for further input is output along with the proposed amount. Further inputs may be provided, for example, via buttons (“Publish” or “Draft”) in hardware (physical buttons, knobs, sliders, etc.), software (GUI, TUI, voice commands, etc.), or any combination thereof. ) and/or via a slider, knob, or similar element.

[0058] Shown in Figure 3 in the same field as the suggested amount is a non-limiting example of a GUI-based slider element that the user can adjust to You can change the suggested selling price. In response to further input, for example, in some embodiments, the proposed amount and/or any of the amounts described above, including any of the at least two amounts that can be used to define any range. You can update and output as well to update any existing output. Any number of buttons or text fields can also be used for input.

[0059] Further amounts are displayed on either side of the central proposed amount in “Proposed Price” such as “Sell Fast” and “Sell Slow” suggestions, which are determined based on the probability scores of the amounts described in 508 below. can do. This combination of suggested prices forms an example of a pricing guide. In other embodiments, other formats and offers may be presented in any configuration to create an equivalent pricing guide, or other guides that use other parameters in addition to or in addition to the listing price. can be created.

[0060] As shown in FIG. 3, additional amounts that can be output via screen display 300 include, for example, the lowest and highest possible prices that can be determined by a given platform. A fee or commission calculation result based on a commission rate or commission rate and a desired selling price that can be selected via elements of screen display 300 may also be displayed. In addition to the calculated fees or commissions, the remaining profits from expected sales may also be displayed.

[0061] Any number of text fields or images may be displayed for additional output. Output can also be achieved by various other means, including sound, tactile feedback, external electronic indicators including light emitting diodes (LEDs), or additional external displays, etc. Any number of output and/or input prompts may be displayed simultaneously or sequentially in any arrangement or timing.

[0062] Although the screen display 300 of FIG. 3 depicts one example configuration, in practice developers and users may use interfaces and devices or systems that implement the enhanced techniques described herein. Any other configurations to assist in designing, implementing, configuring, and using the are possible within the scope and spirit of this disclosure.

[0063] FIG. 4 is a flowchart illustrating a method 400 that includes probabilistic item matching and searching, according to some embodiments. Method 400 is performed by processing logic that can include hardware (e.g., circuitry, dedicated logic, programmable logic, microcode, etc.), software (e.g., instructions executed on a processing device), or a combination thereof. be able to. In all cases, not all steps of method 400 may be necessary to implement the enhanced techniques disclosed herein. Additionally, as will be understood by those skilled in the art, some steps of method 400 may be performed simultaneously or in a different order than shown in FIG.

[0064] Method 400 shall be described with reference to FIGS. 1 and 6. However, method 400 is not limited only to those example embodiments. The steps of method 400 may be performed by at least one computer processor coupled to at least one memory device. Exemplary processors and memory devices are described below with respect to 604 of FIG. In some embodiments, method 400 can be performed using system 100 of FIG. 1, which can further include at least one processor and memory, such as that of FIG. 6.

[0065] At 402, at least one processor, such as processor 604, can load data points associated with a particular item. For example, a data point may be an input value from a user, which may include, for example, a string of letters, numbers, or other information that may potentially identify a particular item. Any loaded data points or received inputs may be stored in memory, such as main memory 608, for example. User input or data points related to particular items may sometimes be incomplete or ambiguous. According to some embodiments, the enhanced techniques for probabilistic matching and searching described herein are utilized to obtain appropriate results despite few, incomplete, or ambiguous inputs. By providing search results, it is possible to improve the accuracy of search results and reduce the level of effort and interaction required of the user.

[0066] In a further example of 402, the data point may be a description, title, name, or a brief characterization or sentence describing the item, which in some embodiments may be optionally entered by the user. I can do it. Depending on the user and the category to which the item belongs, the input may be in response to a particular prompt (not shown), etc. Data points may additionally or alternatively be non-textual data from, for example, sensors, cameras, voice recognition, artificial intelligence, or neural networks, to generate descriptions based on other inputs or environmental factors, etc. can be based on. In some embodiments, data points may be generated, sent and received programmatically and/or in an automated manner, such as by an application programming interface (API), and not necessarily by manual input from a user. .

[0067] At 404, processor 604 can generate a database query based on the data points. Database queries are not limited to traditional structured database queries. Rather, for purposes of this disclosure, a database query is any query that can function like a database query. For example, any term, expression, or value (or any portion thereof) used to determine a match with other data may be considered a database query for purposes of this disclosure. Similarly, in some embodiments, any entry in the Database refers to any listing, data, or item data or corresponding metadata for which the item data or corresponding metadata can be extracted, scraped, parsed, or otherwise processed. May correspond to a structure, web page, or other entity.

[0068] In some embodiments, the query may be a string literal, such as corresponding to a title or description, or the like. In some embodiments, the query may be a regular expression or similar input with special characters or components intended to match more than the literal input string itself. In some embodiments, a database query may include operators and/or syntax such as, for example, an SQL query. However, such queries are not necessarily limited to accessing SQL databases; they can use other types of databases, data stores, data lakes, data pools, data feeds, data streams, etc. Moreover, it does not have to be completely structured, it may not be structured at all, or it may be partially structured. In some embodiments, the query may be used to search the web using, for example, public resources, third party resources, libraries, databases, or web search engines.

[0069] At 406, the processor 604 can receive a response to the database query, the response including a plurality of candidate items related to the particular item. In some embodiments, the received response may be stored in memory, such as memory 608. In some cases, it may be possible that the response has less than one candidate item, i.e. one or zero, but in such cases probabilistic search and matching may be omitted in such cases. , can be terminated early. In other cases where the response to the database query includes results for multiple candidate items, processing of method 400 can continue to determine, eg, with a certain degree of confidence, at least one top result.

[0070] At 408, processor 604 can receive a first input related to a particular item. The first input can be used, for example, to further qualify the data points, regardless of their type or format. In some embodiments, even though the data point is a text object, the first input can be a different text object (e.g., a string), and the data point and the first input are, for example, a text description. can describe different elements (e.g., name, make, model, year, size, etc.) or can be different parts of the same element.

[0071] In other cases, the first input may be of a different type, for example, where the data point may be text in the form of a string, the first input may be a photographic image; or vice versa. Other types can be used, including, for example, voice recognition or other distinctive sounds or voice fingerprints. In addition to text-based strings, mathematical or cryptographic signatures in any representation can also be used. Bar codes or other patterns or sequences can be used.

[0072] Other possible types of input include, but are not limited to, vibration patterns, chemical samples and their analysis, measurements of radiation, electrical and/or magnetic signals, or any other environmental sensor input. Not limited. Other forms of input can be received, such as output from other computer programs or algorithms, such as in the form of neural network output, perceptron output, image recognition output, classification output, or other types of output. These outputs may be based on other user inputs, such as photo data, voice or audio data, or other third party resources, feeds, etc.

[0073] The first input may be mobile, for example, by other visible or audible indicators, via a network, and/or via an API of a program (local or remote on the device performing the probabilistic matching and search). The second input, the third input, etc. may be received in response to any of a variety of prompts, such as on a screen of a device, etc. , or any further input.

[0074] At 410, processor 604 can generate probability scores for at least two candidate items of the plurality of candidate items in the response based on at least the first input. The probability score may be, for example, in absolute terms or relative to other candidates. For terms relative to other candidates, the other candidate may be a candidate received in response to a database query, and/or multiple queries and responses thereto (e.g., at least a predetermined amount of time). (e.g., the history of past queries and responses within the frame). In this latter case of probability score generation from multiple queries or responses, the probability scoring may be cumulative over a particular time period and/or over an entire time period. For such cumulative probability scoring, additional algorithms involving machine learning can, in some embodiments, It may be useful to improve the accuracy of probability scores.

[0075] The probability scores described herein are calculated locally on a client device, e.g., a PC, or a mobile device such as a smartphone or tablet computer, on a self-hosted service (including a dedicated server, VPS, or on-premises cloud). It can be generated remotely on the infrastructure, remotely via a third party service, or any combination thereof. In embodiments where probability score generation can be performed via multiple devices, a first device performs structured data queries (e.g., of markup, serialized formats, key-value pairs, etc.) or other may be configured to transmit or pass a data structure to a second device and initiate or continue processing to generate a resulting probability score, e.g., the first device, the second device module, or a third device. In some embodiments, not all components of a query or data structure may include structured data for the desired operation.

[0076] According to some embodiments, item listings or search results can be biased or ranked based on probability scores that can be generated as described herein. In this way, scoring and ranking implementations may rely on remote databases or third-party service providers to generate matches and/or probability scores and return results in some use cases. can provide a relatively high level of specificity in the results without requiring user input of specificity, such as SKU, on a local device such as a smartphone or PC terminal.

[0077] At 412, processor 604 may select a selected item from the plurality of candidate items based on the selected item's probability score. Based on at least one of the data points and the first input, processor 604 can select the candidate item as the selected item. In some embodiments, processor 604 may select, for example, the candidate item with the highest corresponding probability score as the selected item. Additionally or alternatively, processor 604 may select selected items based on any other criteria or conditions in other embodiments.

[0078] In some embodiments, in response to loading any data points or receiving any first input, second input, third input, or any further input, , the probability score may be updated or otherwise changed. In response, such changes can also be reflected in the selected item to the extent that the candidate item with the highest corresponding probability score can also be changed. For example, processor 604 may change the selected item to the candidate item with the highest corresponding probability score if the selected item is no longer the same as the candidate item with the highest corresponding probability score. In other embodiments, other criteria for selecting, updating, or otherwise modifying selected items may additionally or alternatively be used.

[0079] At 414, processor 604 can output a reference to the selected item. Specifically, as an output, processor 604 may return a reference or some other value as a result in some embodiments. The reference or value may enable a separate function, module, device, user, or other entity to recognize or identify the resulting selection item. The reference or value, in some embodiments, is a uniform, generic, etc., including, but not limited to, SKU, UPC, URI, URL, URN, ISBN, ASIN, etc., to which a given item may be mapped. and/or a unique identifier. Additionally or alternatively, the value may include a checksum, fingerprint, Can include a signature, digest, hash, or cryptographic hash.

[0080] Additionally, the output of 414, either by reference or by value, can be used as an input to further methods, systems, and/or devices. For example, reference or identification values corresponding to selected items may be reliably used as input for further recommendations and/or biasing of search results, such as in method 500 and 502 of FIG. 5, described in more detail below. can be used. Further details of such uses and data flows are further described in the application entitled "Probabilistic Search Biasing and Recommendations" (U.S. Application No. 16/288,373), which is incorporated herein by reference. ing. 414, "Inventory Ingestion, Image Processing, and Market Descriptor Pricing System", "Inventory Ingestion, and Pricing System” and “System and Method 16/288,199, 16/288,203, and 16/288,158, respectively). can.

[0081] Method 400 is disclosed in the order shown above in this exemplary embodiment of FIG. However, in practice, the operations disclosed above may be performed sequentially in any order, along with other operations, or alternatively, performed in parallel, such that two or more operations are performed simultaneously. or any combination thereof.

[0082] FIG. 5 is a flowchart illustrating a method 500 that includes probabilistic item matching and searching, according to some embodiments. Method 500 is performed by processing logic that can include hardware (e.g., circuitry, dedicated logic, programmable logic, microcode, etc.), software (e.g., instructions executed on a processing device), or a combination thereof. be able to. In all cases, not all steps of method 500 may be necessary to implement the enhanced techniques disclosed herein. Additionally, as will be understood by those skilled in the art, some steps of method 500 may be performed simultaneously or in a different order than shown in FIG. 5.

[0083] Method 500 shall be described with reference to FIGS. 1 and 6. However, method 400 is not limited to only those example embodiments. The steps of method 400 may be performed by at least one computer processor coupled to at least one memory device. Exemplary processors and memory devices are described below with respect to 604 of FIG. In some embodiments, method 400 may be performed using system 100 of FIG. 1, which may further include at least one processor and memory, such as that of FIG. 6.

[0084] At 502, a processor, such as processor 604, can receive input related to the identified item. The identified item may, in some embodiments, be the selected item output such as from 414 with respect to FIG. 4 above, which is also incorporated by reference herein. No. 16/288,379, entitled "Probabilistic Item Matching and Searching," published by J.D. As long as the method 500 can utilize any additional or alternative methods of identifying items in addition to or in place of the probabilistic item matching and/or searching described with respect to the method 400 above, it may be received at 502. The input received may be, for example, similar to the input received at 402 above.

[0085] Alternatively, in other embodiments of 502, the input is, for example, a description, a title, a name, or a brief characterization or sentence describing the item, which may be optionally entered by the user. be able to. Depending on the user and the category to which the item belongs, the input may be in response to a particular prompt (not shown), etc. The input may additionally or alternatively be based on non-textual data, for example from a sensor, camera, voice recognition, artificial intelligence, or neural network, to generate a description based on other inputs or environmental factors, etc. be able to. Input can be generated, sent and received programmatically and/or in an automated manner, such as by, for example, an application programming interface (API), and not necessarily by manual input from a user.

[0086] At 504, processor 604 can generate a database query based on the input. Database queries are not limited to traditional structured database queries. Rather, for purposes of this disclosure, a database query is any query that can function like a database query. For example, any term, expression, or value (or any portion thereof) used to determine a match with other data may be considered a database query for purposes of this disclosure. Similarly, in some embodiments, any entry in the Database refers to any listing, data, or item data or corresponding metadata for which the item data or corresponding metadata can be extracted, scraped, parsed, or otherwise processed. May correspond to a structure, web page, or other entity.

[0087] In some embodiments, the query may be a string literal, such as corresponding to a title or description, or the like. In some embodiments, the query may be a regular expression or similar input with special characters or components intended to match more than the literal input string itself. In some embodiments, a database query may include operators and/or syntax such as, for example, an SQL query. However, such queries are not necessarily limited to accessing SQL databases; they can use other types of databases, data stores, data lakes, data pools, data feeds, data streams, etc. Moreover, it does not have to be completely unstructured, it can be completely unstructured, or it can be semi-structured. In some embodiments, the query may be used to search the web using, for example, public resources, third party resources, libraries, databases, or web search engines.

[0088] At 506, the processor 604 can receive a response to the database query, the response includes a plurality of equivalent items, the equivalent item is similar to the identified item, and the response includes a plurality of equivalent items. further including corresponding metadata for a plurality of equivalent items including a range of amounts corresponding to. In some cases, it may be possible that a response has less than one, i.e., one or zero, equivalent items, but in such cases probabilistic recommendations may be omitted in such cases and the early can be terminated. In other cases where the response to the database query includes results for multiple equivalent items, processing of method 500 may continue.

[0089] As a separate but related consideration, if a metadata field of particular interest is blank, erroneous, or otherwise uninformative with respect to a given equivalent item, processor 604: In some cases, corresponding equivalent items may be ignored for further calculations regarding metadata fields of the same interest. In one example use case, a given item may be determined to be an equivalent item by having at least some elements of its description similar to corresponding elements of the input at 502. can. However, according to some embodiments, if, for example, processor 604 can compare comparable items to determine a reasonable price range at which an input item (e.g., that of 502) can sell in a given market, , the processor 604 may ignore certain comparable items if their corresponding price (meta) data is missing or is a statistical outlier (data that is likely to be useless or erroneous). I can do it.

[0090] At 508, the processor 604 can generate probability scores for at least two amounts of the range of amounts based at least on corresponding metadata of the plurality of comparable items. Following the example immediately above, processor 604 can compare comparable items to determine a reasonable price range at which the input item (eg, that of 502) can sell in a given market. A meaningful output may not be a single monetary amount, but may be reflected by two or more monetary amounts, depending on other factors. Additionally or alternatively, two of the at least two amounts may represent a lower amount and a higher amount, for example, defining a range of given amounts.

[0091] For example, the two amounts define a range of prices for which a given item (e.g., entry 502) has at least a certain probability of selling within a given time (period and/or specific date). be able to. Apart from this, different price ranges can be determined for the probability that the sale will be completed regardless of time. In further embodiments, the first amount may represent the highest price at a given probability of sale, independent of time, and the second amount may represent the highest price at a given probability of sale within a given time constraint. It can be determined as the highest price with sales probability.

[0092] At 510, the processor 604 can output at least the suggested amount, along with a prompt for further input, based at least on the probability scores generated for at least two amounts of the range of amounts. The proposed amount, in some embodiments, can be one of the at least two amounts described above with respect to 508. In other embodiments, the proposed amount can be different from any of the at least two amounts at 508, for example, but may in some cases be derived from any of the at least two amounts described above with respect to 508. I can do it.

[0093] For further explanation, an example is shown in FIG. In the example shown in FIG. 3, $990 is a "proposed price" that represents an example of the proposed amount output on the screen display. Additionally, a prompt for further input is output along with the proposed amount. Further input can be, for example, a button (“Submit” or “Save Draft”) in hardware (physical buttons, knobs, sliders, etc.), software (GUI, TUI, voice commands, etc.), or any combination thereof. ”) and/or via a slider, knob, or similar element.

[0094] Shown in Figure 3 in the same field as Offer Amount is a non-limiting example of a GUI-based slider element that the user can adjust to You can change the suggested selling price. In response to further input, for example, in some embodiments, the proposed amount and/or any of the amounts described above, including any of the at least two amounts that can be used to define any range. You can update and output as well to update any existing output.

[0095] Further amounts are displayed on either side of the central proposed amount of “Proposed Price” such as “Sell Fast” and “Sell Slow” suggestions, which are determined based on the probability score of the amount described in 508 above. can do. This combination of suggested prices forms an example of a pricing guide. In other embodiments, other formats and offers may be presented in any configuration to create an equivalent pricing guide, or other guides that use other parameters in addition to or in addition to the listing price. can be created.

[0096] In further embodiments, a time value may be output based on at least one equivalent item and corresponding metadata. Additionally or alternatively, the time value may be set by the user at his or her discretion. Time values that may be stored in memory 608 may be used to calculate, for example, any of the at least two monetary amounts described at 508, at least with any time-dependent probability score. As a further example, a time value may be a period of time for which a sale has a given probability score, such as 6 hours, 2 days, 1 week, etc. Although such time values may be stored and associated with corresponding probability scores or other values, the time values may optionally be displayed to the user, e.g. together with or in place of any other values. You can also do that.

[0097] Similar to the probability score, suggested amount, and other values described above, in some embodiments, the probability score, suggested amount, and other values may be determined in response to any user input (e.g., first input, second input, etc.) or further user input. The updated time values and/or probability scores can be output to update any existing output. Additional operations of updating or refreshing, such as based on any user input and/or any periodic refresh period, may be performed according to any other relevant techniques understood in the art.

[0098] Method 500 is disclosed in the order shown above in this exemplary embodiment of FIG. However, in practice, the operations disclosed above may be performed sequentially in any order, along with other operations, or alternatively, performed in parallel, such that two or more operations are performed simultaneously. or any combination thereof.

exemplary computer system
[0099] Various embodiments may be implemented using one or more computer systems, such as computer system 600 shown in FIG. 6, for example. One or more computer systems 600 can be used to implement, for example, any of the embodiments discussed herein, as well as combinations and subcombinations thereof.

[0100] Computer system 600 may include one or more processors (also referred to as central processing units or CPUs), such as processor 604. Processor 604 may be connected to a bus or communication infrastructure 606.

[0101] Computer system 600 may also include user input/output devices 603, such as a monitor, keyboard, pointing device, etc., which may communicate with communication infrastructure 606 via user input/output interface 602. .

[0102] One or more of processors 604 may be a graphics processing unit (GPU). In one embodiment, a GPU may be a processor, which is a specialized electronic circuit designed to process mathematically intensive applications. GPUs can be used, for example, in computer graphics applications, image, video, vector processing, array processing, etc., as well as cryptography (including brute force cracking), generation of cryptographic hashes or hash sequences, partial hash function inverse problems ( Parallelization of large blocks of data, such as mathematically intensive data, common in solving partial hash-inversion problems) and/or producing the results of other proof-of-work calculations for some blockchain-based applications. It can have a parallel structure that is efficient for processing. The capabilities of general purpose computing with graphics processing units (GPGPUs) make GPUs very useful, at least in the image recognition and machine learning aspects described herein.

[0103] Additionally, one or more of the processors 604 may implement a coprocessor or other logic to accelerate cryptographic calculations or other specialized mathematical functions, including a hardware-accelerated cryptographic coprocessor. can include. Such accelerated processors may further include an instruction set for acceleration using a coprocessor and/or other logic to facilitate such acceleration.

[0104] Computer system 600 may also include main or primary memory 608, such as random access memory (RAM). Main memory 608 may include one or more levels of cache. Main memory 608 may store control logic (ie, computer software) and/or data therein.

[0105] Computer system 600 may also include one or more secondary storage devices or secondary memory 610. Secondary memory 610 may include, for example, a main storage drive 612 and/or a removable storage device or drive 614. Main storage drive 612 may be, for example, a hard disk drive or a solid state drive. Removable storage drive 614 may be a floppy disk drive, magnetic tape drive, compact disk drive, optical storage device, tape backup device, and/or any other storage device/drive.

[0106] Removable storage drive 614 can interact with removable storage unit 618. Removable storage unit 618 may include a computer usable or readable storage device on which computer software (control logic) and/or data is stored. Removable storage unit 618 may be a floppy disk, magnetic tape, compact disk, DVD, optical storage disk, and/or any other computer data storage device. Removable storage drive 614 can read from and write to removable storage unit 618 .

[0107] Secondary memory 610 may include any other means, device, component, equipment, or other approach for allowing computer programs and/or other instructions and/or data to be accessed by computer system 600. can include. Such means, devices, components, equipment, or other approaches may include, for example, removable storage unit 622 and interface 620. Examples of removable storage units 622 and interfaces 620 include program cartridges and cartridge interfaces (e.g., such as those found in video game devices), removable memory chips (e.g., EPROM or PROM) and associated sockets, memory sticks and USB ports. , a memory card and associated memory card slot, and/or any other removable storage unit and associated interface.

[0108] Computer system 600 may further include a communications or network interface 624. Communication interface 624 may enable computer system 600 to communicate and interact with any combination of external devices, external networks, external entities, etc. (individually and collectively referred to by reference numeral 628). For example, communication interface 624 may enable computer system 600 to communicate with an external or remote device 628 via communication path 626, where communication path 626 can be wired and/or wireless (or a combination thereof). ) and can include any combination of LAN, WAN, Internet, etc. Control logic and/or data may be transferred to and from computer system 600 via communication path 626 .

[0109] Computer system 600 may also include a personal digital assistant (PDA), desktop workstation, laptop or notebook computer, netbook, tablet, smartphone, smart watch or other computer, to name a few non-limiting examples. wearables, appliances, parts of the Internet of Things (IoT), and/or embedded systems, or any combination thereof.

[0110] It is to be understood that the frameworks described herein can be implemented as a method, process, apparatus, system, or article of manufacture, such as a non-transitory computer-readable medium or device. For purposes of explanation, the framework may be described in the context of a distributed ledger that is publicly available, or at least available to untrusted third parties. One example of a modern use case is with blockchain-based systems. However, the Framework is also applicable to other situations where sensitive or confidential information may need to be passed by or through the hands of untrusted third parties, and that the present technology is never a distributed ledger or block It should be understood that the application is not limited to chains.

[0111] Computer system 600 may be implemented in a remote or distributed cloud computing solution, local or on-premises software (e.g., an "on-premises" cloud-based solution), or an "as-a-service" model (e.g., content-as-a-service). CaaS), Digital Content as a Service (DCaaS), Software as a Service (SaaS), Managed Software as a Service (MSaaS), Platform as a Service (PaaS), Desktop as a Service (DaaS), Frames as a Service Backend as a Service (FaaS), Backend as a Service (BaaS), Mobile Backend as a Service (MBaaS), Infrastructure as a Service (IaaS), Database as a Service (DBaaS), etc.), and/or the aforementioned examples. Clients or servers that access or host any applications and/or data through any delivery paradigm, including but not limited to hybrid models, including any combination of can do.

[0112] Any applicable data structures, file formats, and schemas may include JavaScript Object Notation (JSON), Extensible Markup Language (XML), another markup language (YAML), Extensible Hypertext Markup Language ( XHTML), Wireless Markup Language (WML), MessagePack, XML User Interface Language (XUL), or other functionally similar representations, alone or in combination. . Alternatively, proprietary data structures, formats, or schemas may be used exclusively or in combination with known or open standards.

[0113] Any associated data, files, and/or databases may be in various human-readable formats, such as numerical, textual, graphical, or multimedia formats, among other possible formats. information may be stored, retrieved, accessed, and/or transmitted, such as in any type of markup language. Alternatively, or in combination with the above formats, data, files, and/or databases may be in binary, encoded, compressed, and/or encrypted format, or any other machine-readable format. The data may be stored, retrieved, accessed, and/or transmitted in any format.

[0114] The interfaces or interconnections between the various systems and layers may be implemented using any number of mechanisms, such as, but not limited to, any number of protocols, programmatic frameworks, floor plans, or application programming interfaces (APIs). Document Object Model (DOM), Discovery Service (DS), NSUserDefaults, Web Services Description Language (WSDL), Message Exchange Patterns (MEP), Web Distributed Data Exchange (WDDX), Web Hypertext Application Technology Working Group (WHATWG) HTML5 Web Messaging, Expression State Transfer (REST or RESTful Web Services), Extensible User Interface Protocol (XUP), Simple Object Access Protocol (SOAP), XML Schema Definition (XSD), XML Remote Procedure Call (XML-RPC), or Any other mechanism, open or proprietary, that can achieve similar functionality and results can be used.

[0115] Such an interface or interconnect may also utilize a Uniform Resource Identifier (URI), which may further include a Uniform Resource Locator (URL) or a Uniform Resource Name (URN). Other forms of uniform and/or unique identifiers, locators, or names may be used, either exclusively or in combination with such forms.

[0116] Any of the protocols or APIs described above can be interfaced with or implemented in any procedural, functional, or object-oriented programming language, and can be compiled or interpreted. Non-limiting examples include C, C++, C#, Objective-C, Java, Scala, Clojure, Elixir, Swift, Go, Perl, PHP, Python, Ruby, JavaScript. ), WebAssembly, or virtually any other language, among many other non-limiting examples. js, V8, Knockout, jQuery, Dojo (registered trademark), Dijit, OpenUI5, AngularJS, Express. js, Backbone. js, Ember. Any other library or library in any kind of framework, runtime environment, virtual machine, interpreter, stack, engine, or similar mechanism, including but not limited to js, DHTMLX, Vue, React, Electron, etc. Include with schema.

[0117] In some embodiments, a tangible non-transitory device or article of manufacture that includes a tangible non-transitory computer-usable or readable medium having control logic (software) stored thereon is described herein. Sometimes referred to in literature as a computer program product or program storage device. This includes, but is not limited to, computer system 600, main memory 608, secondary memory 610, removable storage units 618 and 622, and tangible articles of manufacture embodying any combination of the foregoing. Such control logic, when executed by one or more data processing devices (e.g., computer system 600), can cause such data processing devices to operate as described herein. .

[0118] Methods of making and using embodiments of the present disclosure using data processing devices, computer systems, and/or computer architectures other than those illustrated in FIG. 6 based on the teachings contained in this disclosure include: It will be clear to those skilled in the relevant art. In particular, embodiments may operate using software, hardware, and/or operating system implementations other than those described herein.

conclusion
[0119] It is to be understood that the Detailed Description section, but not all other sections, is intended to be used to interpret the claims. Other sections may describe one or more, but not all, exemplary embodiments contemplated by the inventors and therefore in no way limit the scope of this disclosure or the appended claims. is not intended.

[0120] Although this disclosure describes example embodiments of example fields and applications, it is to be understood that this disclosure is not limited thereto. Other embodiments and modifications thereof are possible and within the scope and spirit of this disclosure. For example, and without limiting the generality of this paragraph, embodiments are not limited to the software, hardware, firmware, and/or entities illustrated and/or described herein. Moreover, embodiments (whether or not explicitly described herein) have significant utility in fields and applications other than the examples described herein.

[0121] Embodiments are described herein using functional building blocks that illustrate the implementation of specified functions and their relationships. The boundaries of these functional building blocks are arbitrarily defined herein for convenience of explanation. Alternative boundaries may be defined so long as the specified functions and relationships (or their equivalents) are properly performed. Also, alternative embodiments may perform functional blocks, steps, acts, methods, etc. using a different order than described herein.

[0122] References herein to "one embodiment," "embodiment," "an exemplary embodiment," "some embodiments," or similar phrases refer to specific embodiments of the described embodiment. may include a particular feature, structure, or characteristic, but does not necessarily indicate that all embodiments may include that particular feature, structure, or characteristic. Moreover, such phrases are not necessarily referring to the same embodiment. Furthermore, when a particular feature, structure, or characteristic is described in connection with one embodiment, the incorporation of such feature, structure, or characteristic into other embodiments is not explicitly mentioned herein. Whether or not mentioned or explained, it would be within the knowledge of those skilled in the relevant art.

[0123] Additionally, some embodiments may be described using the expressions "coupled" and "connected" and derivatives thereof. These terms are not necessarily intended to be synonyms for each other. For example, some embodiments use the terms "connected" and/or "coupled" to indicate that two or more elements are in direct physical or electrical contact with each other. may be used to explain. However, the term "coupled" can also mean that two or more elements are not in direct contact with each other, yet still cooperate or interact with each other.

[0124] The breadth and scope of this disclosure should not be limited by any of the above-described exemplary embodiments, but should be defined only in accordance with the following claims and their equivalents.

[0116] Any of the protocols or APIs described above can be interfaced with or implemented in any procedural, functional, or object-oriented programming language, and can be compiled or interpreted. Non-limiting examples include C, C++, C#, Objective -C, Java, Scala, Clojure, Elixir, Swift, Go, Perl , PHP, Python, Ruby, JavaScript. ) , WebAssembly, or virtually any other language, among many other non-limiting examples. js, V8, Knockout, jQuery, Dojo (registered trademark) , Dijit, OpenUI5, AngularJS, Express. js, Backbone. js, Ember. Any other library or library in any kind of framework, runtime environment, virtual machine, interpreter, stack, engine, or similar mechanism, including but not limited to js, DHTMLX, Vue, React, Electron, etc. Include with schema.

Claims (20)

loading data points related to a particular item by at least one computer processor;
generating, by the at least one computer processor, a database query based on the data points;
receiving, by the at least one computer processor, a response to the database query, the response including a plurality of candidate items related to the particular item;
receiving, by the at least one computer processor, a first input related to the particular item;
generating, by the at least one computer processor, probability scores for at least two candidate items of the plurality of candidate items in the response based at least on the first input;
selecting, by the at least one computer processor, a selected item from the plurality of candidate items based on the probability score of the selected item;
outputting, by the at least one computer processor, a reference to the selected item, the selected item being mapped to a unique identifier;
computer-implemented methods, including; receiving, by the at least one computer processor, a second input related to the particular item;
updating, by the at least one computer processor, the probability score of at least one candidate item of the plurality of candidate items based at least on the second input;
2. The method of claim 1, further comprising: changing, by the at least one computer processor, the selected item to correspond to a different candidate item based at least on the updating the probability score of the at least one candidate item of the plurality of candidate items; And,
outputting the modified selection item by the at least one processor;
3. The method of claim 2, further comprising: 2. The method of claim 1, wherein at least one of the response, the data point, or the first input includes a classification result, an image recognition result, or a combination thereof. 2. The method of claim 1, wherein the response to the database query includes at least one data entry obtained from at least one public source or at least one third party source. memory and
at least one processor coupled to the memory;
, the at least one processor comprising:
receiving data points related to a particular item;
generating a database query based on the data points;
receiving a response to the database query, the response including a plurality of candidate items related to the particular item;
receiving a first input related to the particular item;
generating probability scores for at least two candidate items of the plurality of candidate items in the response based at least on the first input;
selecting a selected item from the plurality of candidate items based on the probability score of the selected item;
outputting a reference to the selected item, the selected item being mapped to a unique identifier;
A system configured to: The at least one processor includes:
receiving a second input related to the particular item;
updating the probability score of at least one candidate item of the plurality of candidate items based on at least the second input;
7. The system of claim 6, further configured to perform. The at least one processor includes:
changing the selected item to correspond to a different candidate item based at least on the updated probability score of the at least one candidate item of the plurality of candidate items;
outputting the changed selected item;
8. The system of claim 7, further configured to perform. 7. The system of claim 6, wherein at least one of the response, the data point, or the first input includes a classification result, an image recognition result, or a combination thereof. 7. The system of claim 6, wherein the response to the database query includes at least one data entry obtained from at least one public source or at least one third party source. A non-transitory computer-readable storage medium having instructions stored thereon, the instructions, when executed by at least one computing device, causing the at least one computing device to perform an operation, the operation comprising:
receiving data points related to a particular item;
generating a database query based on the data points;
receiving a response to the database query, the response including a plurality of candidate items related to the particular item;
receiving a first input related to the particular item;
generating probability scores for at least two candidate items of the plurality of candidate items in the response based at least on the first input;
selecting a selected item from the plurality of candidate items based on the probability score of the selected item;
outputting a reference to the selected item, the selected item being mapped to a unique identifier;
non-transitory computer-readable storage media, including: The said operation is
receiving a second input related to the particular item;
updating the probability score of at least one candidate item of the plurality of candidate items based at least on the second input;
12. The non-transitory computer readable storage medium of claim 11, further comprising: The said operation is
changing the selected item to correspond to a different candidate item based at least on the updating the probability score of the at least one candidate item of the plurality of candidate items;
outputting the changed selected item;
13. The non-transitory computer readable storage medium of claim 12, further comprising: 12. The non-transitory computer-readable storage medium of claim 11, wherein at least one of the response, the data point, or the first input comprises a classification result, an image recognition result, or a combination thereof. A non-transitory computer-readable storage medium having instructions stored thereon, the instructions, when executed by at least one computing device, causing the at least one computing device to perform an operation, the operation comprising:
receiving input related to the identified item;
generating a database query based on the input;
receiving a response to the database query, the response including a plurality of equivalent items, the plurality of equivalent items being similar to the identified item, and the response including the plurality of equivalent items; further including and receiving corresponding metadata for the plurality of equivalent items, including an amount range corresponding to the item;
generating probability scores for at least two amounts in the amount range based at least on the corresponding metadata of the plurality of equivalent items;
outputting at least a suggested amount with a prompt for further input based at least on the generating the probability score for the at least two amounts in the range of amounts;
non-transitory computer-readable storage media, including: The said operation is
16. Based at least on the generating the probability score for the at least two amounts in the range of amounts, the method further comprises outputting at least a lower amount and a higher amount along with the prompt for further input. Non-transitory computer readable storage medium as described. The said operation is
receiving the further input in response to the prompt for the further input;
updating at least one of the time value and the probability score of the at least two amounts of the range of amounts based at least on the further input;
outputting at least the updated probability score for one of the at least two amounts in the range of amounts;
16. The non-transitory computer readable storage medium of claim 15, further comprising: The said operation is
further comprising outputting a time value based at least on the corresponding metadata of the plurality of equivalent items and the generating the probability score for the at least two amounts in the range of amounts. 16. The non-transitory computer readable storage medium of item 15. The said operation is
receiving the further input in response to the prompt for the further input;
updating the probability score of the at least two amounts in the range of amounts based on at least the further input;
outputting at least one of the time value and the updated probability score of one of the at least two amounts of the range of amounts;
20. The non-transitory computer readable storage medium of claim 18, further comprising: The said operation is
16. The non-transitory computer-readable storage medium of claim 15, further comprising modifying a search ranking of the identified item based on at least one metadata entry.