Loan filing information automatic checking and prompting method and device, equipment and medium

Abstract

The invention discloses a loan filing information automatic checking and prompting method and device, equipment and a medium. The method is characterized by comprising the following steps: receiving remotely input loan audio and video information, and determining a dialogue text according to the loan audio and video information; obtaining loan filing information corresponding to the loan audio and video information, and determining at least one piece of abnormal filing information according to the loan filing information and the dialogue text; and determining a risk prompt video clip corresponding to each piece of abnormal filing information according to each piece of abnormal filing information and the loan audio and video information. The dialogue information in the audio and video is automatically extracted, content analysis is carried out according to the loan filing information and the dialogue text, whether the loan filing information has a problem is judged, risk reminding is carried out in time for the place with the problem, and the video clip at the corresponding position with the risk is automatically intercepted and processed, so that rechecking personnel can conveniently and quickly recheck. The cost reduction and efficiency improvement of manpower are realized, the rechecking period is shortened, and the loan risk is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of loan review, and in particular to a method, device, equipment and medium for automatically checking and prompting loan file information. Background Art

[0002] In the bank loan business, the review of key information in the audio and video conversations depends on manual frame-by-frame viewing of archived audio and video to understand the relevant information of the person being investigated, and then check and verify the information with the bank loan business records item by item to determine whether the information is filled in accurately, so as to realize the review of the investigator. The purely manual review method in the existing technology relies too much on manual frame-by-frame viewing and item-by-item comparison. Faced with thousands of pieces of information to be reviewed, the labor cost is high and the review cycle is long. Moreover, people are prone to fatigue due to doing a lot of repetitive work, which leads to the risk of missing some wrong information review. Summary of the invention

[0003] The present invention provides a method, device, equipment and medium for automatically checking and prompting loan file information, so as to solve the risks of slow information review and omission of investigators in loan business in the prior art.

[0004] According to one aspect of the present invention, a method for automatically checking and prompting loan file information is provided, comprising:

[0005] Receiving remotely inputted loan audio and video information, and determining a dialogue text according to the loan audio and video information;

[0006] Acquire loan file creation information corresponding to the loan audio and video information, and determine at least one abnormal file creation information according to the loan file creation information and the conversation text;

[0007] Determine the risk warning video clip corresponding to each abnormal file creation information based on each abnormal file creation information and the loan audio and video information.

[0008] According to another aspect of the present invention, there is provided a device for automatically checking and prompting loan file information, comprising:

[0009] An audio and video processing module, used to receive remotely inputted loan audio and video information, and determine a dialogue text according to the loan audio and video information;

[0010] An abnormality identification module, used to obtain the loan file information corresponding to the loan audio and video information, and determine at least one abnormal file information according to the loan file information and the conversation text;

[0011] The risk warning module is used to determine the risk warning video clip corresponding to each abnormal file creation information according to each abnormal file creation information and the loan audio and video information.

[0012] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0013] at least one processor; and

[0014] a memory communicatively connected to the at least one processor; wherein,

[0015] The memory stores a computer program that can be executed by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the automatic verification and prompt method for loan file information described in any embodiment of the present invention.

[0016] According to another aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the automatic verification and prompting method for loan file creation information described in any embodiment of the present invention when executed.

[0017] The technical solution of the embodiment of the present invention, by receiving the remotely inputted loan audio and video information, determining the dialogue text according to the loan audio and video information, can automatically identify the dialogue text of the loan audio and video information, without manually processing the audio and video information, thereby improving the review efficiency and accuracy; obtaining the loan file information corresponding to the loan audio and video information, determining at least one abnormal file information according to the loan file information and the dialogue text, can automatically perform abnormal processing and positioning on the abnormal file information, thereby improving the recognition accuracy of the abnormal file information; determining the risk warning video clip corresponding to each abnormal file information according to each abnormal file information and the loan audio and video information. Automatically intercepting and processing the risk video clip, convenient for rapid review and positioning, solves the risk of slow review of the information of the investigator of the loan business and omission of the review in the prior art, by automatically extracting the dialogue information in the audio and video, and comparing the loan file information and the dialogue text for content analysis, it is determined whether there is a problem with the loan file information, and timely reminds the risk of the problematic areas, and automatically intercepts and processes the video clip corresponding to the risk source, so that the review personnel can quickly review. It can reduce manpower costs and improve efficiency, shorten review cycles, and reduce loan risks.

[0018] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0020] Figure 1 A flowchart of a method for automatically checking and prompting loan file information is provided for an embodiment of the present invention;

[0021] Figure 2 A flowchart of another method for automatically checking and prompting loan file information provided by an embodiment of the present invention;

[0022] Figure 3 A schematic diagram of the structure of a device for automatically checking and reminding loan file information provided by an embodiment of the present invention;

[0023] Figure 4 A schematic structural diagram of an electronic device 10 that can be used to implement an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0024] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0025] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0026] Figure 1The present invention provides a flowchart of a method for automatically checking and prompting loan file information. This embodiment is applicable to the case of conducting audio and video review of the loan file information of the investigator. The method can be executed by an automatic loan file information checking and prompting device. The automatic loan file information checking and prompting device can be implemented in the form of hardware and / or software. The automatic loan file information checking and prompting device can be configured in an electronic device. Figure 1 As shown, the method includes:

[0027] S110, receiving remotely inputted loan audio and video information, and determining a conversation text according to the loan audio and video information.

[0028] Among them, the loan audio and video information can be the communication audio and video information between the loan staff and the person under investigation. It should be noted that when the person under investigation requests a loan from a bank or other loan issuer, he or she will create loan file information at the loan issuer and fill in the corresponding information for each question in the loan file information; after the person under investigation has filled in the corresponding information, the loan issuer will contact the person under investigation through the loan staff to confirm the loan file information. For example, the loan file information can be that the person under investigation fills in the orchard as collateral in the mortgage of the loan file information, the loan staff contacts the person under investigation to enter the orchard, shoots a video of the orchard, and communicates with the person under investigation about the planting conditions and income information of the orchard, thereby obtaining the loan audio and video information.

[0029] The conversation text may be text information corresponding to the voice conversation in the loan audio and video information. It should be noted that the loan staff communicated with the person under investigation about the loan file information, and the loan audio and video information recorded the conversation voice. The conversation text is obtained by recognizing the conversation voice in the loan audio and video information.

[0030] Optionally, in another optional embodiment of the present invention, determining the dialogue text according to the loan audio and video information includes:

[0031] The audio of the loan audio and video information is extracted by a multimedia video interpretation method to determine the loan audio data; the loan audio data is subjected to noise reduction processing to determine the conversation voice; the conversation voice data is converted into text to determine the conversation text.

[0032] The multimedia video analysis method can be used to analyze the loan audio and video information to obtain the audio information and video information of the loan audio and video information. It should be noted that the loan audio and video information is analyzed by the multimedia video analysis method to obtain the audio information and video information constituting the loan audio and video information, and the audio information is extracted as the loan audio information. Exemplarily, the multimedia video analysis method can include video editing software and an online audio and video conversion website.

[0033] The loan audio data may be audio information in the loan audio and video information. Optionally, the loan audio and video information is a video recording of the conversation between the loan staff and the person under investigation, and the loan audio information is mainly composed of the conversation voice between the loan staff and the person under investigation.

[0034] Optionally, an embodiment of the present invention provides an audio extraction method: each Packet in the loan audio and video information is decoded into a corresponding data stream (video or audio), and when the Packet is read as an audio Packet, the audio Packet is used as the output target and the audio Packet is used as the output file to obtain the loan audio data.

[0035] Among them, the conversation voice can be the conversation voice between the loan staff and the person under investigation in the loan audio information; it should be noted that although the loan audio information is mainly composed of the conversation voice between the loan staff and the person under investigation, there will be certain noise and background sound in the recording process of the loan audio and video information, and the loan audio information needs to be denoised to obtain the conversation voice.

[0036] Optionally, after obtaining the loan audio information, the background noise of the loan audio information is identified and removed by setting a noise reduction tool, and the loan audio information is subjected to noise reduction processing by setting noise reduction parameters to obtain a conversation voice.

[0037] Optionally, in the embodiment of the present invention, the text conversion of the conversational speech can be performed by automatic speech recognition. Automatic Speech Recognition (ASR) is a speech-to-text technology that can convert the conversational speech into a conversational text. Specifically, the speech is first located by endpoint detection, the start and end points of the speech are located, and the silent part and the noise part are removed to find the truly effective content of the conversational speech. Then, the conversational speech is denoised by an audio denoising method, and then equalized and normalized, and the conversational speech is feature extracted, and cut into small segments at very short time intervals to form frames. For each frame, the Mel spectrum feature of the audio is then extracted and converted into a multidimensional vector; the conversational speech is decoded with the help of the maximum matching probability of the pre-established acoustic model and the prediction result of the language model, and the multidimensional vector is converted into text, and output as a conversational text. Among them, the pre-established acoustic model processes the encoded vectors, combines adjacent frames into phonemes, and then combines them into single words or Chinese characters; the language model is used to adjust the illogical words obtained by the acoustic model to make the recognition results smooth.

[0038] Optionally, a speech library for training the acoustic model is pre-built, and audio features are extracted based on the training speech data in the speech library to obtain training audio features, and then the acoustic model is trained; a text library for training the acoustic model is built, and text features are extracted based on the training text data in the text library to obtain training text features, and then the language model is trained.

[0039] Specifically, audio is extracted from the loan audio and video information through a multimedia video interpretation method to determine the loan audio data; noise reduction is performed on the loan audio data to determine the conversation voice; and text conversion is performed on the conversation voice data to determine the conversation text.

[0040] S120. Obtain loan file information corresponding to the loan audio and video information, and determine at least one abnormal file information according to the loan file information and the conversation text.

[0041] The loan file information may be the loan information filled in by the person under investigation when requesting a loan from the loan issuer. The loan file information is in a unified format and consists of two parts: one is the target survey question and the other is the investigator's loan information. The target survey question may be the question that the loan issuer collects information from the loan requester (the person under investigation), and the investigator's loan information may be the respondent's response to the target survey question.

[0042] Among them, abnormal file creation information may be information that there is inconsistency between the investigator's loan information of the person under investigation and the information answered by the person under investigation in the loan audio and video information. It should be noted that when the loan staff verifies the investigator's loan information, the information answered by the person under investigation is required to be consistent with the investigator's loan information filled in. If the two are inconsistent, the inconsistent investigator's loan information will be regarded as abnormal file creation information.

[0043] Specifically, for the person under investigation, the loan file information corresponding to the loan audio and video information is obtained, the conversation text of the person under investigation is compared with the loan file information, and at least one abnormal file information in the loan file information is determined.

[0044] S130. Determine the risk warning video clip corresponding to each abnormal file creation information according to each abnormal file creation information and the loan audio and video information.

[0045] Among them, the risk warning video clip can be a video clip captured from the loan audio and video information; the risk warning video clip can be used to remind the loan provider that there are risks in the information of the person under investigation during the actual investigation process.

[0046] Specifically, after obtaining the abnormal filing information, the loan audio and video information is cut based on the abnormal filing information to obtain a risk warning video clip corresponding to each abnormal filing information.

[0047] The technical solution of the embodiment of the present invention, by receiving the remotely inputted loan audio and video information, determining the dialogue text according to the loan audio and video information, can automatically identify the dialogue text of the loan audio and video information, without manually processing the audio and video information, thereby improving the review efficiency and accuracy; obtaining the loan file information corresponding to the loan audio and video information, determining at least one abnormal file information according to the loan file information and the dialogue text, can automatically perform abnormal processing and positioning on the abnormal file information, thereby improving the recognition accuracy of the abnormal file information; determining the risk warning video clip corresponding to each abnormal file information according to each abnormal file information and the loan audio and video information. Automatically intercepting and processing the risk video clip, convenient for rapid review and positioning, solves the risk of slow review of the information of the investigator of the loan business and omission of the review in the prior art, by automatically extracting the dialogue information in the audio and video, and comparing the loan file information and the dialogue text for content analysis, it is determined whether there is a problem with the loan file information, and timely reminds the risk of the problematic areas, and automatically intercepts and processes the video clip corresponding to the risk source, so that the review personnel can quickly review. It can reduce manpower costs and improve efficiency, shorten review cycles, and reduce loan risks.

[0048] Figure 2This is a flow chart of another automatic verification and prompting method for loan file information provided by an embodiment of the present invention. The relationship between this embodiment and the above embodiment is a specific method for identifying abnormal file information through dialogue text. Figure 2 As shown, the method includes:

[0049] S210: Receive remotely inputted loan audio and video information, and determine a conversation text according to the loan audio and video information.

[0050] S220: Decompose the structured information of the loan file creation information to determine at least one target survey question and the investigator loan information corresponding to each target survey question.

[0051] Optionally, since the loan file information is a fixed structured file, the target survey questions of the loan file information and the investigator's loan information corresponding to each target survey question can be directly identified through structured information decomposition.

[0052] Optionally, the steps for decomposing structured information of loan file information in an embodiment of the present invention are as follows: performing layout detection on the loan file information, identifying structural elements in the loan file information, and determining the spatial coordinates and arrangement order of the structural elements, extracting text information of the loan file information through optical character recognition, and restoring the logical relationship and order of the text information based on the spatial coordinates and arrangement order to obtain each target survey question and the investigator loan information corresponding to each target survey question.

[0053] S230: Performing text semantic similarity analysis on the dialogue text according to the target survey questions in turn, and determining loan identification information corresponding to each target survey question.

[0054] The loan identification information may be the voice response information of the person being investigated to the target investigation question identified in the conversation text. It should be noted that when the loan staff conducts an on-site investigation of the person being investigated, they will ask each target investigation question in the loan file information and fully record the voice response of the person being investigated.

[0055] Optionally, the text semantic similarity analysis of the target survey question and the dialogue text can be performed by using a pre-trained deep model of text semantic similarity, the target survey question and the dialogue text are input into the deep model, and the deep model outputs the loan identification information corresponding to the target survey question. The training process of the deep model of text semantic similarity is as follows: a large amount of text data in the text library is used to train a pre-trained language model (PLM) to obtain a pre-trained language model with strong representation, in preparation for subsequent text feature representation. Then, a large number of text pairs with similar text semantics and text pairs with dissimilar text semantics are sorted, and data enhancement is performed through back translation, mask, synonym replacement and other methods to obtain richer text pairs. The pre-trained model is fine-tuned using the enhanced data. Then, some features are randomly removed through the pooling layer to improve the generalization of the model. Finally, in order to improve the comparison ability of the model similarity, a contrastive learning method is added to make similar text features closer and dissimilar text pairs farther apart, so as to improve the semantic similarity ability of the model and obtain a deep model of text semantic similarity.

[0056] Specifically, the embodiment of the present invention sequentially inputs the target survey questions and the dialogue text into a deep model of text semantic similarity to perform text semantic similarity analysis, and obtains that the deep model of text semantic similarity analysis outputs the loan identification information corresponding to each target survey question.

[0057] Optionally, in another optional embodiment of the present invention, the step of performing text semantic similarity analysis on the dialogue text according to the target survey questions in turn to determine the loan identification information corresponding to each target survey question includes:

[0058] Perform text semantic similarity analysis on the target survey question and the dialogue text in turn to locate the answer text corresponding to each target survey question; identify the answer text corresponding to each target survey question through named entity recognition to determine the loan identification information.

[0059] The answer text may be the text information in which the respondent responds to the target survey question in the dialogue text. It should be noted that since the loan staff usually asks each target survey question in the loan file information of the respondent separately, there is an answer text corresponding to each target survey question in the dialogue text.

[0060] Optionally, when performing text semantic similarity analysis on the target survey questions and the dialogue text, since the dialogue text includes the text of the loan staff's target survey questions to the respondents, the deep model of text semantic similarity performs similarity matching on the target survey questions in the dialogue text, obtains the matching text content in the dialogue text, and determines the answer text around the text content based on the semantics of the text content.

[0061] Optionally, after obtaining the answer text of the target survey question, the answer text corresponding to each target survey question is identified through named entity recognition, and then the loan identification information is extracted from the answer text. Among them, the named entity model can be used to perform named entity recognition on the answer text. The named entity model can extract entities with specific meanings in combination with the context content of the answer text, and identify the loan identification information in the entity. It should be noted that the named entity model mainly includes a word segmenter, BERT feature extraction, BiLSTM (Bidirectional Long Short-Term Memory) module and CRF (Conditional Random Field) decoding strategy module. When the named entity model is recognized, the word segmenter is first used to segment the text, and then the discrete one-hot features are mapped into low-dimensional dense features using a pre-trained language model. The mapped features are input into the BiLSTM module, and the attributes of each feature and the relationship with other features are analyzed in combination with the context information of the text. Finally, the named entity is extracted through the CRF decoding strategy to determine the loan identification information.

[0062] Specifically, the embodiment of the present invention sequentially inputs the target survey question and the dialogue text into a deep model of text semantic similarity to perform text semantic similarity analysis, obtains the text content in the dialogue text that matches the target survey question, and then determines the answer text corresponding to the target survey question.

[0063] Optionally, in another optional implementation of the present invention, the performing text semantic similarity analysis according to the target survey question and the dialogue text in sequence to locate the answer text corresponding to each target survey question includes:

[0064] Perform text semantic similarity analysis on the target survey question and the dialogue text in turn to locate the dialogue positioning point corresponding to each target survey question; and determine the answer text corresponding to each target survey question in the dialogue text according to the dialogue positioning point.

[0065] Among them, the dialogue positioning point can be the marking point of the target survey question in the dialogue text; the dialogue positioning point can be used to determine the position of the target survey question in the dialogue text, and then extract the dialogue positioning point corresponding to the target survey question.

[0066] Optionally, when performing text semantic similarity analysis on the target survey question and the dialogue text, since the dialogue text includes the text of the target survey question asked by the loan staff to the person being investigated, the deep model of text semantic similarity performs similarity matching on the target survey question in the dialogue text, obtains the matching dialogue anchor points in the dialogue text, determines the text content of the target survey question in the dialogue text based on the dialogue anchor points, and determines the answer text around the text content based on the semantics of the text content. When extracting the answer text, the dialogue anchor points of the target survey question in the dialogue text and the dialogue anchor points of other target survey questions adjacent to the target survey question can also be used to extract the question, and the text content between the dialogue anchor points of the target survey question and the dialogue anchor points of other target survey questions can be extracted to obtain the answer text.

[0067] Specifically, text semantic similarity analysis is performed on the target survey question and the dialogue text in turn to locate the dialogue positioning point corresponding to each target survey question; and the answer text corresponding to each target survey question is determined in the dialogue text according to the dialogue positioning point.

[0068] The embodiment of the present invention sequentially inputs the target survey question and the dialogue text into a deep model of text semantic similarity to perform text semantic similarity analysis, obtains dialogue anchor points in the dialogue text that match the target survey question, and determines the answer text corresponding to each of the target survey questions in the dialogue text based on the dialogue anchor points.

[0069] Optionally, in another optional embodiment of the present invention, determining the answer text corresponding to each target survey question in the dialogue text according to the dialogue anchor point includes:

[0070] The question answer range corresponding to each of the target survey questions is determined according to each of the dialogue positioning points; and the answer text corresponding to each of the target survey questions is determined in the dialogue text according to the question answer range.

[0071] The scope of the question answer may be the text scope of the respondent's response to the target survey question in the conversation text. It should be noted that when the loan staff records the respondent's voice response, the respondent may have a voice conversation with the loan staff during the voice response, resulting in the target survey question being fragmented and the answer text corresponding to the target survey question cannot be directly extracted.

[0072] Optionally, after obtaining the dialogue positioning point corresponding to the target survey question, the position of each target survey question in the dialogue text is determined through each dialogue positioning point, and then the text content between the target survey question and the previous and next adjacent target survey questions is used as the question answer range of the target survey question, and the answer text corresponding to each target survey question is determined in the dialogue text based on the question answer range corresponding to the target survey question.

[0073] S240, comparing the loan identification information corresponding to the target survey question with the investigator's loan information one by one, and determining at least one abnormal file creation information,

[0074] Specifically, the loan identification information corresponding to each target survey question is compared with the investigator's loan information. If the loan identification information is consistent with the investigator's loan information, the next target survey question is compared; if the loan identification information is inconsistent with the investigator's loan information, the target survey question and the investigator's loan information corresponding to the target survey question are treated as abnormal filing information; until all target survey questions are compared and at least one abnormal filing information is determined.

[0075] S250. Determine the risk warning video clip corresponding to each abnormal file creation information according to each abnormal file creation information and the loan audio and video information.

[0076] Optionally, in another optional embodiment of the present invention, determining the risk warning video clip corresponding to each abnormal file creation information according to each abnormal file creation information and the loan audio and video information includes:

[0077] Determine an abnormal investigation question based on each abnormal file creation information; perform dialogue recognition and video segmentation on the loan audio and video information to determine at least one investigation dialogue segment; match each abnormal investigation question with each investigation dialogue segment at the dialogue positioning point corresponding to the dialogue text to determine the risk warning video segment corresponding to each abnormal file creation information.

[0078] The abnormal investigation question may be a target investigation question corresponding to the abnormal file creation information. The investigation dialogue segment may be an audio or video segment of a voice dialogue between a loan staff and a person under investigation on the target investigation question in the loan audio or video information.

[0079] Optionally, after obtaining the loan audio and video information, the loan audio and video information is subjected to dialogue recognition to determine the video content of each dialogue on the target survey question in the loan audio and video information, and then the loan audio and video information is subjected to video segmentation to determine the survey dialogue segment corresponding to at least one target survey question corresponding to the loan audio and video information. The survey dialogue segment is also set as a timestamp in the loan audio and video to indicate the video time range of the survey dialogue segment in the loan audio and video.

[0080] Optionally, for each abnormal investigation question, obtain the corresponding dialogue positioning point of the abnormal investigation question in the dialogue text, determine the corresponding time range of the abnormal investigation question in the loan audio and video information based on the dialogue positioning point, and then match it with the timestamp of the investigation dialogue segment to obtain the investigation dialogue segment corresponding to each abnormal filing information, and use the investigation dialogue segment as a risk warning video segment for the abnormal filing information.

[0081] The technical solution of the embodiment of the present invention determines the dialogue positioning point of the target investigation question in the dialogue text corresponding to the loan audio and video information through text semantic similarity analysis, identifies the loan identification information based on the dialogue positioning point, and then realizes the review of the loan file information. After the review is completed, the risk video clip can be automatically intercepted and processed based on the dialogue positioning point, which is convenient for rapid review and positioning, and solves the risk of slow review of investigator information and omissions in the existing technology for loan business. By automatically extracting the dialogue information in the audio and video, and comparing the loan file information and the dialogue text for content analysis, it is determined whether there is a problem with the loan file information. For problematic areas, risk reminders are given in a timely manner, and the video clip corresponding to the risk source is automatically intercepted and processed, which is convenient for reviewers to quickly review. It can achieve cost reduction and efficiency improvement of manpower, reduce the review cycle, and reduce the risk of loans.

[0082] Figure 3 The present invention provides a schematic diagram of a structure of a device for automatically checking and reminding loan file information according to an embodiment of the present invention. Figure 3 As shown, the device includes: an audio and video processing module 310, an abnormality identification module 320 and a risk warning module 330; wherein,

[0083] The audio and video processing module 310 is used to receive the remotely inputted loan audio and video information and determine the dialogue text according to the loan audio and video information;

[0084] An abnormality identification module 320, used to obtain the loan file information corresponding to the loan audio and video information, and determine at least one abnormal file information according to the loan file information and the conversation text;

[0085] The risk warning module 330 is used to determine the risk warning video clip corresponding to each abnormal file creation information according to each abnormal file creation information and the loan audio and video information.

[0086] The technical solution of the embodiment of the present invention, by receiving the remotely inputted loan audio and video information, determining the dialogue text according to the loan audio and video information, can automatically identify the dialogue text of the loan audio and video information, without manually processing the audio and video information, thereby improving the review efficiency and accuracy; obtaining the loan file information corresponding to the loan audio and video information, determining at least one abnormal file information according to the loan file information and the dialogue text, can automatically perform abnormal processing and positioning on the abnormal file information, thereby improving the recognition accuracy of the abnormal file information; determining the risk warning video clip corresponding to each abnormal file information according to each abnormal file information and the loan audio and video information. Automatically intercepting and processing the risk video clip, convenient for rapid review and positioning, solves the risk of slow review of the information of the investigator of the loan business and omission of the review in the prior art, by automatically extracting the dialogue information in the audio and video, and comparing the loan file information and the dialogue text for content analysis, it is determined whether there is a problem with the loan file information, and timely reminds the risk of the problematic areas, and automatically intercepts and processes the video clip corresponding to the risk source, so that the review personnel can quickly review. It can reduce manpower costs and improve efficiency, shorten review cycles, and reduce loan risks.

[0087] Optionally, the abnormality identification module 320 is specifically used for:

[0088] Decomposing the structured information of the loan file creation information, determining at least one target survey question and the investigator's loan information corresponding to each target survey question;

[0089] Performing text semantic similarity analysis on the dialogue texts according to the target survey questions in turn, and determining loan identification information corresponding to each target survey question;

[0090] The loan identification information corresponding to the target investigation question and the investigator's loan information are compared one by one to determine at least one abnormal filing information.

[0091] Optionally, the abnormality identification module 320 is further configured to:

[0092] Performing text semantic similarity analysis on the target survey question and the dialogue text in turn to locate the answer text corresponding to each target survey question;

[0093] The answer text corresponding to each target survey question is identified through named entity recognition to determine the loan identification information.

[0094] Optionally, the abnormality identification module 320 is further configured to:

[0095] Performing text semantic similarity analysis on the target survey question and the dialogue text in turn, and locating the dialogue positioning point corresponding to each target survey question;

[0096] The answer text corresponding to each target survey question is determined in the dialogue text according to the dialogue positioning point.

[0097] Optionally, the abnormality identification module 320 is further configured to:

[0098] Determine the question answer range corresponding to each target survey question according to each dialogue positioning point;

[0099] The answer text corresponding to each of the target survey questions is determined in the dialogue text according to the question answer range.

[0100] Optionally, the risk warning module 330 is specifically used for:

[0101] Determine an abnormal investigation question based on each abnormal file creation information;

[0102] Performing dialogue recognition and video segmentation on the loan audio and video information to determine at least one investigation dialogue segment;

[0103] According to each of the abnormal investigation questions, the dialogue positioning point corresponding to the dialogue text is matched with each of the investigation dialogue segments, and the risk warning video segment corresponding to each of the abnormal filing information is determined.

[0104] Optionally, the audio and video processing module 310 is specifically used for:

[0105] Performing audio extraction on the loan audio and video information by a multimedia video interpretation method to determine the loan audio data;

[0106] Performing noise reduction processing on the loan audio data to determine the conversation voice;

[0107] The conversation voice data is converted into text to determine the conversation text.

[0108] The automatic verification and prompting device for loan file creation information provided in the embodiment of the present invention can execute the automatic verification and prompting method for loan file creation information provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0109] Figure 4A schematic diagram of the structure of an electronic device 10 that can be used to implement an embodiment of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their modes are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0110] like Figure 4 As shown, the electronic device 10 includes at least one processor 11, and a memory connected to the at least one processor 11, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 to the random access memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other through a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0111] A number of components in the electronic device 10 are connected to the I / O interface 15, including: an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices through a computer grid such as the Internet and / or various telecommunication grids.

[0112] The processor 11 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 11 executes the various methods and processes described above, such as the automatic verification and prompting method for loan file creation information.

[0113] In some embodiments, the automatic verification and prompting method for loan file information can be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed on the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the automatic verification and prompting method for loan file information described above can be performed. Alternatively, in other embodiments, the processor 11 can be configured to execute the automatic verification and prompting method for loan file information in any other appropriate manner (for example, by means of firmware).

[0114] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0115] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer program is executed by the processor, the modes / operations specified in the flow chart and / or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0116] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in combination with an instruction execution system, device or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0117] To provide interaction with a user, the systems and techniques described herein may be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0118] The systems and techniques described herein can be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a grid browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication grid). Examples of communication grids include: a local area network (LAN), a wide area network (WAN), a blockchain grid, and the Internet.

[0119] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication grid. The client and server relationship is generated by computer programs running on the respective computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0120] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0121] This embodiment provides a computer-readable storage medium having a computer program stored thereon. When the program is executed by a processor, the steps of the automatic verification and prompting method for loan file creation information provided in any embodiment of the present invention are implemented. The method includes:

[0122] Receiving remotely inputted loan audio and video information, and determining a dialogue text according to the loan audio and video information;

[0123] Acquire loan file creation information corresponding to the loan audio and video information, and determine at least one abnormal file creation information according to the loan file creation information and the conversation text;

[0124] Determine the risk warning video clip corresponding to each abnormal file creation information based on each abnormal file creation information and the loan audio and video information.

[0125] The computer storage medium of the embodiment of the present invention can adopt any combination of one or more computer-readable media. The computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium can be, for example, but not limited to: an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples (non-exhaustive list) of computer-readable storage media include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this document, a computer-readable storage medium can be any tangible medium containing or storing a program, which can be used by an instruction execution system, device or device or used in combination with it.

[0126] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, which carry computer-readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer-readable signal media may also be any computer-readable medium other than a computer-readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0127] The program code embodied on the computer readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

[0128] Computer program code for performing the operation of the present invention may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of grid, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0129] It should be understood by those skilled in the art that the above modules or steps of the present invention can be implemented by a general-purpose computing device, they can be concentrated on a single computing device, or distributed on a grid composed of multiple computing devices, optionally, they can be implemented by a program code executable by a computer device, so that they can be stored in a storage device and executed by the computing device, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. Thus, the present invention is not limited to any specific combination of hardware and software.

[0130] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0131] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for automatically checking and reminding loan file information, characterized in that: include: Receiving remotely inputted loan audio and video information, and determining a dialogue text according to the loan audio and video information; Acquire loan file creation information corresponding to the loan audio and video information, and determine at least one abnormal file creation information according to the loan file creation information and the conversation text; Determine the risk warning video clip corresponding to each abnormal file creation information based on each abnormal file creation information and the loan audio and video information.

2. The method according to claim 1, characterized in that The determining of abnormal file creation information according to the loan file creation information and the conversation text includes: Decomposing the structured information of the loan file creation information, determining at least one target survey question and the investigator's loan information corresponding to each target survey question; Performing text semantic similarity analysis on the dialogue texts according to the target survey questions in turn, and determining loan identification information corresponding to each target survey question; The loan identification information corresponding to the target investigation question and the investigator's loan information are compared one by one to determine at least one abnormal filing information.

3. The method according to claim 2, characterized in that The step of performing text semantic similarity analysis on the dialogue texts according to the target survey questions in turn to determine the loan identification information corresponding to each target survey question includes: Performing text semantic similarity analysis on the target survey question and the dialogue text in turn to locate the answer text corresponding to each target survey question; The answer text corresponding to each target survey question is identified through named entity recognition to determine the loan identification information.

4. The method according to claim 3, characterized in that The step of performing text semantic similarity analysis on the target survey question and the dialogue text in turn to locate the answer text corresponding to each target survey question includes: Performing text semantic similarity analysis on the target survey question and the dialogue text in turn, and locating the dialogue positioning point corresponding to each target survey question; The answer text corresponding to each target survey question is determined in the dialogue text according to the dialogue positioning point.

5. The method according to claim 4, characterized in that The step of determining the answer text corresponding to each target survey question in the dialogue text according to the dialogue positioning point includes: Determine the question answer range corresponding to each target survey question according to each dialogue positioning point; The answer text corresponding to each of the target survey questions is determined in the dialogue text according to the question answer range.

6. The method according to claim 5, characterized in that The step of determining the risk warning video clip corresponding to each abnormal file creation information according to each abnormal file creation information and the loan audio and video information includes: Determine an abnormal investigation question based on each abnormal file creation information; Performing dialogue recognition and video segmentation on the loan audio and video information to determine at least one investigation dialogue segment; According to each of the abnormal investigation questions, the dialogue positioning point corresponding to the dialogue text is matched with each of the investigation dialogue segments, and the risk warning video segment corresponding to each of the abnormal filing information is determined.

7. The method according to claim 1, characterized in that Determining the dialogue text according to the loan audio and video information includes: Performing audio extraction on the loan audio and video information by a multimedia video interpretation method to determine the loan audio data; Performing noise reduction processing on the loan audio data to determine the conversation voice; The conversation voice data is converted into text to determine the conversation text.

8. A device for automatically checking and reminding loan file information, characterized in that: include: An audio and video processing module, used to receive remotely inputted loan audio and video information, and determine a dialogue text according to the loan audio and video information; An abnormality identification module, used to obtain the loan file information corresponding to the loan audio and video information, and determine at least one abnormal file information according to the loan file information and the conversation text; The risk warning module is used to determine the risk warning video clip corresponding to each abnormal file creation information according to each abnormal file creation information and the loan audio and video information.

9. An electronic device, characterized in that: The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the automatic verification and prompt method for loan file creation information as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the automatic verification and prompting method for loan file creation information according to any one of claims 1 to 7 when executed.

Images