FR3103937A1 - IDENTIFICATION AND UNDERSTANDING OF COMPUTER-CONTROLLED QUESTIONS IN A COMMERCIAL TENDER FILE FOR AUTOMATED TENDER PROCESSING - Google Patents

Abstract

In a method of identifying and understanding computer-driven issues in a commercial tender dossier (CTD), different CTDs from different people are uploaded to fixed storage on a computer. Then, the different CTDs are loaded into memory and grouped together in different groupings according to at least one grouping criterion. For each of the CTDs, a corresponding grouping among the groupings is identified and a segmentation model is selected chosen from among the identified groupings. Subsequently, segmentation is performed on the CTD using the selected segmentation model to produce a set of the segmented portions of the CTD. Finally, for each of the segmented portions, a set of extraction rules mapped to said portion is retrieved, at least one question extracted from said segmented portion using the set of extraction rules and each extracted question stored in a database to productive participation of questions for the CPCs. Figures for the abstract: Figure 1

Description

IDENTIFICATION AND UNDERSTANDING OF COMPUTER-DRIVEN QUESTIONS IN A COMMERCIAL TENDER DOCUMENT FOR AUTOMATED TENDER PROCESSING

Technical field of the invention

The present invention relates to the field of the processing of commercial offers and more particularly the automated processing of commercial tender files.

Technical background

Commercial bidding documents refer to the set of basic business documents used in commerce for the acquisition of goods or services. Commercial tender documents range from a request for tender (RFT) to an invitation to bid (ITB) when the nature of the goods or services to be acquired is well understood, at the request for proposal (RFP for "request for proposal" in English) when the nature of the goods or services to be acquired is less clear. In all cases, the essential structure of commercial bidding documents includes a series of questions that the prospective respondent must answer. In some cases, the questions are organized in a structurally reasonable way. But, in other cases, the questions lack adequate organization.

The process for responding to a commercial tender involves identifying each question in a commercial tender document and formulating a written response. The written response, therefore, may be compared with other written responses so that the requester can award one of the bidders a subsequent supply contract, or so that the requester can offer one or more bidders a request for subsequent offer such as a request for information (RFI for "request for information" in English) or a price request (RFQ for "request for quotation" in English). In either case, the exercise of responding to a commercial RFP tends to be a manual exercise that is repeated for each commercial RFP.

Recognizing the repetitive nature of responding to a commercial RFP, software vendors in the past have offered automated and semi-automated tools to facilitate the generation of a response to a commercial RFP. Typical solutions focus on the RFP response process and rely on a centralized corporate information repository from which standardized responses to RFP questions can be selected and inserted into a response template. RFP, either manually, automatically or, in most cases, a combination of both. Some solutions allow responses to be tagged with metadata to help organize pre-built response content, easily accessible by the end user. In fact, some solutions go so far as to classify the freshness or timeliness of response content to ensure that the most recent response content is used in an RFP response.

However, in all cases, RFP response management solutions focus only on a repository of responses to questions used in the past without regard to the context of the questions being asked and more specifically, the context of the RFP itself. Obviously, the best answer to a question depends on the context of the RFP. However, modern automated RFP solutions do not take into account the RFP context. Additionally, while the question sets in an RFP may look different, resulting in an automated solution that provides a mapping of responses to the specific question being presented, many questions may be semantically identical despite using different words. . In this case, a wide range of otherwise viable responses will be excluded from use in selecting an appropriate response. Finally, although modern RFP response management tools select responses presented in past RFP responses, no consideration is provided as to the individual probability of success of each selected response.

The embodiments of the present invention fill the gaps in the state of the art with regard to the automated production of a response document for a commercial tender document (CTD for "commercial tender document" in English ) and provide a novel and non-obvious method, system, and computer program product for identifying and understanding computer-based issues in a CTD. In a method of identifying and understanding computer-driven issues in a CTD, different CTDs of different people are downloaded to fixed storage of a computer from a computer communication network from different computer clients. Then, the different CTDs among the CTDs are loaded into the memory of the computer. The different CTDs can then be grouped into multiple different groupings according to at least one grouping criterion.

For each of the CTDs, a corresponding one of the clusters is identified and a segmentation model is selected for the identified one of the clusters. For example, the segmentation model can be a model of a prototype CTD for industrial classification indicating different topical sections of the prototype CTD. Subsequently, segmentation is performed on the CTD of CTDs using the selected segmentation template to produce a set of segmented portions of said CTD of CTDs. Finally, for each segmented portion, a set of extraction rules mapped to the segmented portion is retrieved, at least one question is extracted from the segmented portion using the extraction rule set, and each extracted question is stored in a crowd-sourced database of questions for CTDs.

In one aspect of the embodiment of the invention, the grouping criterion is an industry classification for a different source of each of the CTDs. Additionally, in another embodiment of the invention, the industry classification is specified in the machine-readable metadata included with each of the CTDs. In another embodiment of the invention, the industry classification is determined by analyzing keywords from each of the CTDs and for each of the CTDs, the analyzed keywords are matched to words known to match an industry classification. specific. In yet another aspect of the embodiment of the invention, the industrial classification is determined by analyzing keywords from each of the CTDs and for each of the CTDs, by subjecting the corresponding keywords among the analyzed keywords to a neural network trained to correlate submitted terms with a particular industry classification.

Finally, another aspect of the embodiment of the invention also includes natural language processing of each question extracted from the crowdsourcing database. The questions are then grouped in the database based on similarity based on natural language processing. Then, a simplest form of the grouped questions is selected as a representative question among the questions, and the grouped questions are matched with the simplest form of the grouped questions. Finally, the simplest form of the grouped questions is stored in the database as a representative question among the grouped questions.

In another embodiment of the invention, a data processing computer system is configured for computer-based identification and understanding of questions in a CTD. The system includes a host computer system that includes one or more computers, each with memory and at least one processor. The system also includes a data memory coupled to the computer system which stores therein a plurality of different CTDs received from a computer communication network from different computer clients of different people. Finally, the system includes a question identification module. The module includes computer program instructions executing in the memory of the host computer system. Program instructions, in turn, make it possible to carry out a process of identifying and understanding questions by computer in a CTD.

Specifically, when executing in memory of the host computer system, the instructions load a plurality of CTDs into memory, and group the CTDs into multiple different groupings according to at least one grouping criterion. For each of the CTDs, a corresponding cluster can be identified from the clusters and a segmentation model can be selected for the cluster identified from the clusters. Then, segmentation can be performed on each CTD using the selected segmentation model to produce a set of segmented portions of the CTD among the CTDs. Finally, for each of the segmented portions, an extraction rule set mapped to the segmented portion is received and at least one question extracted from the segmented portion using the extraction rule set. Then, each extracted question is stored in a crowdsourced database of questions for CTDs.

In this regard, since the crowdsourcing database includes questions for CTDs from different entities, patterns can be identified among the questions. Specifically, since similar questions are grouped under a common genre, the questions that are of most interest in the different CTDs being processed at any given time can be identified as a question pattern. For each processed question extracted from a CTD, the question can be associated with a particular group and a number of questions for each group can be calculated. Therefore, a distribution of questions received in the set can be observed in such a way as to reveal, by calculation, questions with the highest prevalence for a given period of time.

Other aspects of the invention will be set forth in part in the description which follows and in part will be obvious from the description, or may be learned by practice of the invention. The aspects of the invention will be realized and achieved by means of the elements and combinations particularly indicated in the various embodiments described. It is understood that the foregoing general description and the following detailed description are both exemplary and explanatory and are not restrictive of the invention.

Brief description of figures

The accompanying drawings, which are incorporated into and form part of this specification, illustrate embodiments of the invention and serve, together with the description, to explain the principles of the invention. The embodiments illustrated herein are presently preferred, it being understood, however, that the invention is not limited to the specific arrangements and instruments shown, in which:

Figure 1 is a pictorial illustration of a computer-driven question identification and understanding process in a CTD;

Figure 2 is a schematic illustration of a data processing computer system suitable for identifying and understanding computer-driven questions in a CTD,

Figure 3 is a flowchart illustrating a process for identifying and understanding computer-driven issues in a CTD.

Detailed description of the invention

Embodiments of the invention identify and understand computer-driven questions in a document such as a CTD. In accordance with the embodiment of the invention, different CTDs are received from different people and are then grouped into different groupings according to grouping criteria, for example an industrial classification of a source of a corresponding CTD among the CTDs. Subsequently, for each CTD, a corresponding cluster is identified, and a segmentation model is selected for the identified cluster among the clusters. Using the segmentation model the segmentation is performed on the CTD so as to produce a set of segmented portions of the CTD. Finally, for each segmented portion, a mapped extraction rule set is retrieved, at least one question extracted from the segmented portion using the extraction rule set, and each extracted question is stored in a database. Crowd-sourced data questions for CTDs. Therefore, optimal answers can be formulated for each of the questions and automatically included in response to a subsequently received CTD by matching the questions in the subsequent CTD to those in the crowdsourcing database and providing the optimal answers to those questions. .

In another illustration, Figure 1 schematically illustrates a process of identifying and understanding computer-driven questions in a CTD. As shown in Figure 1, a multiplicity of different CTDs 100A, for example RFQs, RFPs, ITBs, RFIs and the like, are received from different people or organizations from a computer communication network such as the global internet. CTD 100As are grouped into different groupings based on grouping criteria, such as associated industry classification, geographic region, or demographic profile, to name a few examples. Once grouped, each grouped CTD 100B is then subjected to a segmentation based on the application of a segmentation model 120 in a pool 130 of segmentation modules 120, each corresponding to a different classification defined by the grouping criteria 110.

The segmentation model 120 may be a model associated with the specific classification, providing an expected architecture for the grouped CTD 100B - namely the number and type of different sections or segments 140A, 140N to be present in the grouped CTD 100B according to the corresponding classification. For example, since the grouped CTD 100B is classified as relating to building construction, segments 140A, 140N would be specified in the segmentation model 120 selected based on material requirements, construction phases, licensure and contractors, whereas to the extent that the aggregated CTD 100B is classified as relating to healthcare services, segments 140A and 140N would be specified in the selected template 120 according to the healthcare providers' biographies, hospital privileges, disease management skills, etc.

Once the aggregated CTD 100B has been segmented into different segments 140A, 140N, a set of one or more extraction rules 160 is selected for each of the segments 140A, 140N. Each of the extraction rules 160 provides a directive to locate a question 150 in a corresponding segment among segments 140A, 140N. Examples include steps including identifying text preceding specific punctuation such as a question mark or colon, or text following specific verbs indicating a question, such as "state", "provide", "list", " indicate", "explain", etc. As each question 150 is extracted from a corresponding one of segments 140A, 140N, question 150 is stored in a question database 170 for further processing. In particular, further processing may include formulating responses to match each of the stored questions 150, the responses of which may be reused in response to any number of CTDs 100A subsequently received.

It should be noted that the question extraction process can be an automated process that depends exclusively on the extraction rules 160. In addition, the question extraction process can be a human-machine hybrid process in which different segments of segments 140A, 140N are manually annotated by an end user through a user interface at CTD 100B. The annotation applied to each 150 question is then associated with the structure of the question so that 150 questions of similar location are assigned the same annotation. Indeed, the extraction rules among the extraction rules 160 can be deduced on the basis of a correlation between the terms of each annotated question among the questions and the annotations themselves. Additionally, once an extraction rule 160 has been inferred from a human annotation of question 150, the extraction rule 160 can be presented in the user interface for confirmation by the end user.

Alternatively, questions 150 from database 170 may be further grouped for semantic similarity. In this regard, each of the 150 questions can be subjected to natural language processing to reduce each of the 150 questions to a simpler form so that the sets of 150 questions can be considered as the same general question when the 150 questions of all have the same semantic result of natural language processing. By grouping 150 questions into the 170 database, each answer already mapped to a genre form of a set of 150 questions can be applied to all 150 question species in the set without requiring separate mapping of each answer. to each species of questions 150 of the set.

The process described in conjunction with Figure 1 can be implemented in a data processing computer system. In another illustration, Figure 2 schematically shows a data processing computer system suitable for identifying and understanding computer-driven questions in a CTD. The system includes a host computer system 200. The host computer system 200 includes one or more computers, each with memory and at least one processor. A CTD response management application 240 runs in memory of the host computing platform 210 and provides automated management of the generation of response documents to the various CTDs. Different end users access the CTD response management application 240 from a computer communication network 210 via different user interfaces 260 provided in different client computing devices 260.

It should be noted that the CTD 240 response management applications facilitate the assembly of a response document to a particular CTD by identifying the questions in the particular CTD, and mapping the identified questions among similar or identical questions stored in the question data memory 220 and the corresponding answers stored in the answer data memory 230. To do this, the question extraction module 300 is coupled to the CTD answer management application 240. During execution within the host computing platform 200, the question extraction module 300 is configured to automate the population of the question data store 220 with questions identified from a multiplicity of CTDs provided by end users via the various user interfaces 260 to generate a crowd-sourced database of questions in data store 220.

Specifically, question extraction module 300 is defined by computer program instructions that are activated during runtime to group together a series of CTDs received over a computer communications network 210 by classifying each of the CTD. Specifically, the program instructions determine the classification of each of the CTDs based on a number of mechanisms. These mechanisms may include the extraction, from each of the CTDs, of metadata indicating the classification. Other mechanisms include parsing the text of each CTD to identify different terms and comparing those terms to the keywords present in a classification table 280A. A predominance of certain terms correlated to keywords of the same classification leads to the assignment of this classification to the CTD. Finally, another mechanism includes parsing the content of each CTD, filtering the content to a subset of terms in the CTD, and submitting the subset of terms to a classification neural network 280A trained to correlate different sets of terms with a probability of a particular classification.

In any case, once the question extraction module program instructions 300 have grouped the CTDs by classification, the program instructions select a segmentation model 290 for each of the CTDs corresponding to the classification assigned to the grouping of the CTD . The program instructions then apply the selected segmentation model 290 to each corresponding CTD of the CTDs to segment each of the CTDs into different segments. Finally, the program instructions, for each segment of each of the CTDs, locate in an extraction rule table 250 a specific set of extraction rules for each segment, and apply the located extraction rules to the respective segment in order to identify and retrieve one or more questions for storage in question data store 220. In this way, question data store 220 can be populated with a selection of crowdsourcing questions which can then be processed individually or by group of questions grouped together to formulate answers in the answer data memory 230. Once the answers are present in the answer data memory 230, the CTD answer management application 240 can facilitate access to these responses, or even automatically populate a response to a concurrently received CTD with responses mapped to recognized questions in the concurrently received CTD and determined to be similar or identical to the corresponding questions in question data store 220.

To further illustrate the operation of question extraction module 300, Figure 3 is a flowchart illustrating a process for identifying and understanding computer-driven questions in a CTD. From block 210 a CTD is selected for processing and in block 320 the CTD text is parsed and passed to a classifier to classify the CTD as belonging to a particular grouping of CTDs. Next, in block 330, a selected segmentation template that matches the cluster classification and the template is applied to the CTD in block 340, to segment the CTD into different segments in which different questions can be found.

In block 350, a first of the segments is selected for processing and in block 360, an extraction rule is located for the segment from a table correlating extraction rules to segments. In block 370, the localized extraction rule is applied to the segment text to extract a question. In box 380, the extracted question can be simplified semantically, for example, by natural language processing or speech identification part, identification of synonymous terms and sentence reduction. In block 390, the simplified question is then compared to existing simplified questions in the question data store to determine if the retrieved question should be grouped with other similar questions belonging to the same genre as the simplified question. If so, the extracted question is grouped under the genre of the simplified question to which one or more answers may already have been mapped in the answer data store.

Finally, in decision block 400, if there remain additional segments to process in the CTD, in block 410 a next segment in the CTD is selected to process the process returns to block 360. Otherwise, the process returns to block 310 with retrieving a next CTD for processing. It is important to note that the above process can continue for each CTD received including CTDs received at the same time so that the question data store is constantly changing in terms of the number and nature of questions therein. are stored, as well as the grouping of questions into different groupings.

The present invention may be embodied in a system, method, computer program, or any combination thereof. The computer program may include a computer-readable storage medium or a medium on which are located computer-readable program instructions to cause a processor to perform aspects of the present invention. The computer-readable storage medium may be a tangible device that can retain and store instructions for use by an instruction executing device. The computer-readable storage medium may be, for example, an electronic storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination thereof. foregoing, but not limited to.

The computer-readable program instructions described herein may be downloaded from a computer-readable storage medium or onto an external computer or external storage device via a network to the respective computing/processing devices. The computer-readable instructions may be executed wholly on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer, and partly on a remote computer or entirely on the remote computer or server. Certain aspects of the present invention are described herein using flow charts and/or block diagrams of methods, apparatus (systems), and computer program products based on embodiments of the invention. It is understood that each block of the flowcharts and/or block diagrams, as well as combinations of blocks of the flowcharts and/or block diagrams, can be realized by means of computer readable program instructions.

These computer readable program instructions may be supplied to a processor of a general purpose computer, special purpose computer or other programmable data processing device to produce a machine, such that the instructions, which are executed through the computer processor or other programmable data processing device, create means to perform the functions/actions specified in the block(s) of the flowcharts and/or block diagrams. These computer-readable program instructions may also be stored in a computer-readable storage medium which may direct a computer, programmable data processing apparatus, and/or other devices to operate in a particular manner, so that the computer-readable storage medium in which instructions are stored comprises an article of manufacture comprising instructions which perform aspects of the functions/actions specified in the block(s) of the flowcharts and/or block diagrams.

The computer-readable program instructions may also be loaded into a computer, other programmable data processing device, or other device to cause a series of operational steps to be performed on the computer, other programmable device, or another device to produce a computer-executed process, such that the instructions that execute on the computer, other programmable device, or other device implement the functions/actions indicated in the block(s) of the flowcharts and/or block diagrams.

The flowchart and block diagrams presented in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products based on the various embodiments of the present invention. In this regard, each block in the flowchart or block diagrams may represent a module, segment or part of instructions, which includes one or more executable instructions to implement the specified logical function(s). In some alternative implementations, the functions noted in the block may occur in the order shown in the figures. For example, two blocks shown in succession may, in fact, be executed substantially simultaneously, or the blocks may sometimes be executed in reverse order, depending on the functionality involved. It should also be noted that each block of the block diagrams and/or flowcharts, and combinations of blocks of the block diagrams and/or flowcharts, may be implemented by special hardware systems that perform the specified functions or acts or execute combinations of special equipment and computer instructions.

Finally, the terminology used herein is only intended to describe particular embodiments and is not intended to limit the invention. It will be further understood that the terms "comprises" and/or "comprising", when used in this specification, indicate the presence of indicated characteristics, integers, steps, operations, elements and/or components, but do not do not exclude the presence or addition of one or more other characteristics, integers, steps, operations, elements, components and/or groups thereof.

The corresponding structures, materials, acts and equivalents of all means or elements of the step and function in the various embodiments of the invention are intended to include any structure, material or act to perform the function in combination with other elements. The description of the present invention has been presented for purposes of illustration and description, but is not intended to be exhaustive or limited to the invention in the form disclosed. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the invention. The embodiment has been selected and described in order to best explain the principles of the invention and its practical application, and to enable others skilled in the art to understand the invention for various embodiments with various suitable modifications for the particular intended use.

Having thus described the invention of the present application in detail and by reference to its embodiments, it will be apparent that modifications and variations are possible without departing from the defined scope of the invention.

Claims (13)

A computer data processing system configured to identify and understand computer-driven issues in a commercial tender document (CTD), the system comprising:
a host computer system comprising one or more computers, each with memory and at least one processor; a data memory coupled to the computer system and storing therein a multiplicity of different CTDs received from a computer communication network from different computer clients, from different people; And,
a question identification module comprising computer program instructions executing in memory of the host computer system, the program instructions performing the steps of:
loading into memory of the host computer system a plurality of CTDs;
grouping the CTDs into multiple different groupings according to at least one grouping criterion; And,
for each of the CTDs:
identifying a matching one of the clusters;
selecting a segmentation model for the identified cluster from the clusters; performing segmentation on said CTD of CTDs using the selected segmentation template to produce a set of segmented portions of said CTD of CTDs; And,
for each of the segmented portions:
retrieving a set of extraction rules mapped to said one of the segmented portions;
extracting at least one question from said one of the segmented portions using the set of extraction rules; And,
store each extracted question in a crowdsourced database of questions for CTDs. A system according to claim 1, wherein the grouping criterion is an industry classification for a different source of each of the CTDs. A system according to claim 2, wherein the industry classification is specified in machine-readable metadata included with each of the CTDs. A system according to claim 2, wherein the industry classification is determined by analyzing keywords from each of the CTDs and for each of the CTDs, by matching the analyzed keywords to words known to match a specific industry classification. The system of claim 2, wherein the industry classification is determined by analyzing keywords from each of the CTDs and for each of the CTDs, subjecting the corresponding ones of the analyzed keywords to a neural network trained to correlate the terms submitted with a particular industrial classification. A system according to claim 2, wherein the segmentation model is a model of a prototype CTD for industrial classification indicating different topical sections of the prototype CTD. A system according to claim 1, wherein the program instructions further perform the steps of:
process in natural language each question extracted from the crowdsourcing database;
group questions in the database based on similarity based on natural language processing;
selecting a simplest form of the grouped questions as a representative question among the questions; And,
associate the grouped questions with the simplest form of the grouped questions and store the simplest form of the grouped questions in the database as the representative question among the questions for the grouped questions. A computer program product for identifying and understanding computer-driven issues in a commercial tender document (CTD), the computer program product comprising program instructions recorded on a computer-readable storage medium computer, the program instructions being executable by a device to cause the device to perform the steps of:
downloading into a fixed storage of a computer from a computer communication network from different computer clients, different CTDs of different people;
loading into computer memory a plurality of CTDs;
grouping the CTDs into multiple different groupings according to at least one grouping criterion; And,
for each of the CTDs:
identifying a matching one of the clusters;
selecting a segmentation model for the identified cluster from the clusters; performing segmentation on said CTD of CTDs using the selected segmentation template to produce a set of segmented portions of said CTD of CTDs; And,
for each of the segmented portions:
retrieving a set of extraction rules mapped to said one of the segmented portions;
extracting at least one question from said one of the segmented portions using the set of extraction rules; And,
store each extracted question in a crowdsourced database of questions for CTDs. A computer program product according to claim 8, wherein the grouping criterion is an industry classification for a different source of each of the CTDs. A computer program product according to claim 9, wherein the industry classification is specified in machine-readable metadata included with each of the CTDs. A computer program product according to claim 9, wherein the industry classification is determined by parsing keywords from each of the CTDs and for each of the CTDs, matching the parsed keywords to words known to match a specific industrial classification. A computer program product according to claim 9, wherein the segmentation model is a model of a prototype CTD for industrial classification indicating different topical sections of the prototype CTD. A computer program product according to claim 8, wherein the steps performed by the device further include the steps of:
process in natural language each question extracted from the crowdsourcing database;
group questions in the database based on similarity based on natural language processing;
selecting a simplest form of the grouped questions as a representative question among the questions; And,
associate the grouped questions with the simplest form of the grouped questions and store the simplest form of the grouped questions in the database as the representative question among the questions for the grouped questions.