Construction of cultural and creative element resource library based on collection information management system

Abstract

In order to improve the effect of collection information management, this paper uses intelligent computer technology to build a cultural and creative element resource library based on the collection information management system. This paper optimizes the design of TPBF by analyzing the false positive rate of the two-stage Bloomfilter query algorithm, determines the use conditions of the new algorithm, and obtains the optimal parameters of the two-stage Bloomfilter query algorithm according to the model. Aiming at the limited improvement of the standard Bloomfilter algorithm and its extended algorithm to reduce the false positive rate, this paper proposes a two-stage Bloomfilter algorithm. Moreover, this paper analyzes the parameters when the algorithm achieves the optimal result, and then proves that the result achieved by the algorithm is consistent with the theoretical expectation through simulation experiments. The experimental research shows that the cultural and creative element resource library based on the collection information management system proposed in this paper can effectively improve the information processing effect of museum collections.

Keywords

collection information management system cultural and creative elements resource library

Introduction

Museum work is complicated and tedious, and the management of museum collections requires patience and meticulousness. The collections in the museum’s collection are very precious, not only of cultural value but also of very important research value. Moreover, a piece of history and culture, even a craft and technology, is covered behind a collection. The protection and research of collections is a kind of inheritance and understanding of history and culture, so for modern people, museum collections have diversified value functions, so they need to be protected and researched. However, the efficiency of manual management is low, the classification and recording of many collections are prone to errors, and the processing and grasping of collection information has not achieved comprehensive and detailed systematic information. This is likely to result in unclear records of collection information, revealing wrong information for historical and cultural research.¹ The rapid development of information technology can just make up for the shortcomings of manual management and recording. The grasp of collections should be refined and rigorous, and attention should be paid to sorting out and mastering the details of the information. Therefore, the creation of an information-based system management environment can fully improve the fineness of museum collection management.²

The complexity and systematicity of museum work far exceed surface recognition, and its core difficulty lies in the multidimensional requirements of collection management. As the material carrier of human civilization, museum collections have three attributes: cultural inheritance, academic research, and artistic aesthetics. Behind each artifact, there are specific historical characteristics, aesthetic orientations, and social contexts.³ This particularity determines that collection management cannot be simply equated with storage style storage of ordinary items, but requires the establishment of a professional management system that integrates cultural relic protection, material science, and historical research. However, traditional management models have structural deficiencies in both methodology and technology: manual registration is not only time-consuming and labor-intensive, but also leads to information gaps due to the lack of standardized processes—according to the International Council of Museums, about 38% of small and medium-sized museums worldwide still use paper catalogs, with 23% of collections having classification errors or age labeling problems.⁴ The distortion of this basic data can create a “butterfly effect” in the research field, and when scholars conduct cultural tracing based on incorrect collection information, it may lead to systematic bias in academic conclusions.⁵

Although the transformation of information technology has partially solved the problem of data storage, existing systems generally suffer from the drawback of fragmented functional modules. Most management systems only realize the entry of structured data such as collection numbers, names, and ages, and lack effective digital analysis tools for unstructured data (such as pattern features, material textures, color lineages, and other cultural and creative core elements). More importantly, there is a serious phenomenon of data silos between systems, with collection databases and cultural and creative development systems often belonging to different platforms, resulting in a broken chain of cultural element transformation from academic research to industrial application. This fragmentation directly restricts the value release of “museum IP”—the practice of the British Museum has shown that systematic management of cultural and creative elements can improve the efficiency of derivative product development by 40%. However, currently 87% of museum cultural and creative works in China are still in the primary stage of simply replicating the appearance of cultural relics.⁶

The intervention of intelligent technology will reconstruct the paradigm of collection management. By using computer vision to automatically extract patterns, constructing semantic correlation networks through natural language processing, and establishing cultural element lineages based on knowledge graphs, these technological means can not only solve the accuracy problems in traditional management, but also activate the value of collections in emerging fields such as digital cultural and creative industries and virtual exhibitions. In particular, the ability of deep learning algorithms to mine hidden cultural features can break through the limitations of manual identification experience. For example, the restoration of Tongjing paintings in the Palace Museum’s Juanqin Studio utilized AI to identify 17 layers of pigment overlay sequences that are difficult for the naked eye to observe. Building such an intelligent management system is essentially rebuilding the “cultural gene bank” of cultural relics in the digital space. Its significance goes beyond improving management efficiency and is a fundamental project for the dynamic inheritance of cultural heritage.⁷

Artificial intelligence technology is gradually becoming the core technical support for collection management. Through deep learning algorithms, the system can automatically recognize cultural relic features and establish semantic association networks, significantly improving the efficiency of collection classification and retrieval. Especially the application of computer vision in pattern extraction and color analysis has broken through the limitations of traditional manual identification. The deterministic and real-time characteristics of artificial intelligence have shown great potential in the scientific management of collections.⁸

The Internet of Things technology has built an intelligent network system for collection management by connecting various sensing devices. RFID technology, as a typical representative, has achieved full lifecycle tracking and management of collectibles, improving the efficiency of inbound and outbound storage by more than 50%. The introduction of environmental monitoring systems has increased the stability of the preservation status of collections by 30%, forming a preventive protection mechanism of “monitoring evaluation warning regulation” ⁹.

Knowledge graph technology provides a semantic knowledge representation method for collection management. By describing entity associations through nodes and relationships, automated reasoning and management of collection knowledge have been achieved. This technology is particularly suitable for visualizing complex collection relationships, providing a new analytical dimension for academic research.¹⁰

High-precision 3D scanning technology can generate millimeter level precision cultural relic models, supporting virtual restoration and stereoscopic display. Through digital collection, a complete digital archive of cultural relics has been established, providing an alternative solution for the protection of fragile cultural relics.¹¹

Collection management has entered the era of big data, and data mining can reveal the historical and cultural information hidden in cultural relics. Intelligent algorithms can analyze massive collection data and discover association patterns that are difficult for traditional research methods to detect. Many institutions have established collection data analysis platforms to support curatorial decision-making and research innovation.¹²

Blockchain provides unique, authentic, and permanent technical guarantees for digital collectibles. In terms of cultural relic tracing and copyright protection, the immutability of blockchain technology has solved the problem of property rights confirmation in traditional management. This technology is driving the standardized development of the digital collectibles market.¹³

Cloud computing technology enables remote storage and collaborative processing of collection data. The digital resources of museums support cross institutional data sharing and research collaboration.¹⁴

Mobile application technology has changed the traditional visiting experience. Through AR technology and intelligent recommendation algorithms, viewers can receive personalized navigation services. The existing intelligent navigation system has increased the efficiency of obtaining cultural relic information by more than 40%.¹⁵

The lack of industry standards remains an important factor restricting the application of technology. At present, the International Council of Museums (ICOM) is promoting the development of a unified metadata standard to enable data exchange between different systems. The National Cultural Heritage Administration is also improving relevant technical specifications.¹⁶

Intelligent management technology is deeply integrated with the cultural and creative industry. Through cross-border application of technology, the value realization path of collections in digital cultural tourism, education and other fields has been expanded. The smart museum has established an innovative model of “culture + technology” to promote the creative transformation of traditional cultural resources.¹⁷

The optimization of TPBF (two-stage Bloomfilter) algorithm is of significant necessity in cultural resource management. Cultural resource information is complex and of great value, and its management requires high-precision and low misjudgment rate. By optimizing the TPBF algorithm, the misjudgment rate can be significantly reduced, query efficiency can be improved, and the accuracy and completeness of cultural resource information can be ensured. This optimization not only improves the speed and accuracy of information retrieval, but also enhances the security and traceability of cultural resources, providing solid technical support for the protection, inheritance, and utilization of cultural resources. Therefore, TPBF optimization is a key link in modernizing and intelligentizing cultural resource management, and is of great significance for promoting the prosperity and development of cultural undertakings.

This paper constructs a resource library of cultural and creative elements based on the collection information management system through intelligent computer technology, and improves the management effect of museum collection information.

Collection information processing algorithm

Design and analysis of TPBF algorithm

In the case of the known set U and set A, the principle of the TPBF algorithm designed in this paper is shown in Figures 1 and 2.

Figure 1.

Creating TPBF from set A and universal set U.

Figure 2.

Creating TPBF from set A and set U-A.

Figure 1 shows the construction process of the two-stage Bloomfilter (TPBF) algorithm: first, initialize two standard Bloomfilters (BF1 and BF2), and insert set A into BF1 through a hash function. Subsequently, BF1 is used to filter the entire set U and collect misclassified elements to form set E. Finally, E is inserted into BF2 to form a TPBF structure consisting of BF1 (main filter) and BF2 (misclassified correction). This figure clearly presents the core mechanism of the algorithm to reduce the false positive rate through processes and annotations.

TPBF is composed of two Bloomfilters. As shown in Figure 1, its representation process is as follows: First, it initializes the bit vectors of two standard Bloomfilters, BF1 and BF2, and sets each bit to 0.

It inserts all elements in set A into BF1, that is, set A is represented by a standard Bloomfilter BF1.

The standard Bloomfilter BF1 is used to query each element in the universal set U, and the set (A + E) is obtained.

Comparing the result of the query with the known set A, we can get all the non-A set elements that pass BF1 in the query, and thus the misjudgment set E is obtained.

Then, the obtained false positive set E is represented by the second standard Bloomfilter BF2. When the elements in set E are inserted into the second Bloomfilter, the creation process of TPBF ends.

In Figures 1 and 2, the set E is the misjudgment set formed by the misjudgment of elements in the set (U-A) by BF1.

If set A and set (U-A) can be obtained from the beginning, then TPBF can also be created as shown in Figure 2.

Figure 2 visually illustrates the process of creating a two-stage Bloomfilter (TPBF) from known sets A and U-A through a flowchart. Firstly, initialize two Bloomfilters (BF1 and BF2), insert the elements of set A into BF1, then use BF1 to query each element in set U-A, collect the elements misclassified as set A to form a misclassified set E, and finally insert the misclassified set E into BF2 to complete the construction of TPBF. This process effectively reduces the false positive rate during the query process.

First, the algorithm initializes the two standard Bloomfilter BF1 and BF2 bit vectors, setting each bit to 0.

The algorithm inserts all elements in set A into BF1, that is, set A is represented by a standard Bloomfilter BF1.

The algorithm uses the standard Bloomfilter BF1 to query each element in the set (U-A), and those elements judged by BF1 as belonging to the set A constitute the misjudgment set E.

The algorithm directly inserts the elements of the set (U-A) that are misjudged as belonging to set A by BF1 into the second standard Bloomfilter BF2, that is, BF2 is used to represent the misjudged set E, and the creation process of TPBF ends here.

The element query process of TPBF is shown in Figure 3.

Figure 3.

The query process of TPBF.

When it is necessary to query whether an element X of the universal set U is a member of the set A, it is possible to perform a two-stage test.

The first stage is the BF1 test. Only when the element X to be checked belongs to the set A or the misjudgment set E will be confirmed by BF1, that is, it is considered to be an element belonging to the set A by BF1. When the element X to be checked belongs to the set (U-A), it will be rejected by BF1. That is, it will be judged by BF1 that it does not belong to the set A, In the first stage of using the TPBF algorithm, when the element X to be checked does not belong to the target set A, how does the standard Bloomfilter BF1 reject X through hash mapping and bit vector judgment, and determine that X does not belong to set A. Therefore, at this stage, most of the elements that do not belong to set A can be clearly determined as not belonging to set A through the test of BF1.

For those to-be-checked elements that cannot be accurately judged whether they belong to the set A or the misjudged set E in the first stage BF1, the second stage test is required to further distinguish. During the second-stage judgment, TPBF uses another standard Bloomfilter BF2 to judge whether the element X to be checked belongs to the set E of misjudgments. If the element judged by BF2 does not belong to the set E, it can be determined that the element belongs to the set A, otherwise the element may belong to the misjudged set E.

Figure 3 shows the query process of a two-stage Bloomfilter (TPBF). When it is necessary to query whether an element X in the entire set U is a member of set A, the first step is to perform preliminary filtering through the standard Bloomfilter BF1 in the first stage. If BF1 determines that X may belong to set A or misjudges set E, the second stage is entered, and another standard Bloomfilter BF2 is used to further determine whether X indeed belongs to misjudged set E. If BF2 is negative, X is determined to be a member of set A; If BF2 confirms, X may belong to the misjudgment set E, but the probability of this situation is extremely low. Through two-stage filtering, the TPBF algorithm achieves highly accurate member queries.

The two-stage Bloomfilter (TPBF) algorithm uses two standard Bloomfilters. When constructing TPBF, a standard Bloomfilter BF1 representing set A is first used for filtering. Here, we set the set $A = {a_{1}, a_{2}, \dots, a_{n}}$ , the number of elements of set A is $n_{A},$ the number of hash functions of $BF 1$ is $k_{1}$ , and the set of elements that $BF 1$ misjudges is $E = {e_{1}, e_{2}, \dots, e_{n_{-} B F A}}$ . Then, the elements in the false positive set of BF1 are inserted into the second Bloomfilter BF2.

In member query, the algorithm firstly uses BF1 to query the element to be checked. If BF1 determines that the element to be checked is not a member of set A, then the element is definitely not a member of set A. If BF1 determines that the element to be checked is a member of set A, the element may or may not be a member of set A. At this time, the non-member element may be misjudged as a member of set A due to the false positive misjudgment of Bloomfilter. Therefore, it is necessary to carry out the judgment of the second stage, and use the BF2 representing the set of misjudgments to query the elements to be checked again. If it is rejected by BF2, the element must not be a member of the misjudged set E, that is, the element must be a member of the set A. Otherwise, the element may be a member of the misjudged set E or a member of the set A, but the probability is quite small. In this way, the sets E and $E^{'} = {{e_{1}}^{'}, {e_{2}}^{'}, \dots, {e_{n_{-}}}^{'} B r_{2}} \in A$ are obtained, as well as the sets $A^{'} = {{a_{1}}^{'}, {a_{2}}^{'}, \dots, {a_{n_{-} {r e g e c t}_{B} B r - 2}}^{'}} \in A$ rejected by BF2, and $A^{'} \cup E^{'} = A$ . As a result, a small amount of misjudgment occurred to the member set A, so the TPBF as a whole showed the properties of zero false positive and low false negative misjudgment rate.

Optimal parameter selection of TPBF

This section optimizes the design of TPBF by analyzing the false positive rate of TPBF, and determines various parameters of TPBF.

If the number of elements in set A is $n_{A}$ , and the number of elements in set U is $n_{U}$ , then the number of non-A set elements is:

n_{U - A} = n_{U} - n_{A}

(1)

The length of the bit vector of the first standard Bloomfilter BF1 is $m_{1}$ , and the number of hash functions of $BF 1$ is $k_{1}$ . Since BF1 is used to represent member set A, the number $n_{1}$ of set elements represented by BF1 is:

n_{1} = n_{A}

(2)

Because BF1 and BF2 are both standard Bloomfilters, there is a possibility of false positives and misjudgments. BF1 will misjudgment some elements of the set U-A as members of the set A, and these elements constitute the misjudgment set $E_{1}$ of BF1. However, BF2 will misjudge some elements in set A as belonging to set $E_{1}$ , thus forming the misjudgment set $E_{2}$ of BF2.

The number of elements of the misjudgment set $E_{1}$ is $n_{F 1}$ , the length of the bit vector of the second standard Bloomfilter BF2 is $m_{2}$ , and the number of hash functions of $BF 2$ is $k_{2}$ . BF2 represents the misjudgment set $E_{1}$ of all elements on BF1, so the number $n_{2}$ of set elements represented by BF2 is:

n_{2} = n_{E : 1}

(3)

Then, the loading factor of BF1 is

{L o a d F a c t o r r}_{i} = m_{1} / n_{1}

(4)

Then the loading factor of BF2 is:

L o a d F a c t o r r_{2} = m_{2} / n_{2}

(5)

The length of TPBF is defined as the sum of the lengths of the bit vectors of the two Bloomfilters, BF1 and BF2. The total length of the bit vectors of TPBF is m, that is,

{m = m}_{1} + m_{2}

(6)

Then the loading factor of TPBF can be obtained as:

{L o a d F a c t o r}_{7 p^{' B} :} = m / n_{A}

(7)

The standard Bloomfilter BF1 represents the set $A = {a_{1}, a_{2}, \dots, a_{n a}}$ , and the false positive misjudgment probability of BF1 is:

P_{B F 1} = {(1 - e^{- k_{i} n_{1} / m_{1}})}^{k_{1}}

(8)

Then the number of false positives and false positives of BF1 for the set (U-A) is:

n_{E 1} = (n_{U} - n_{A}) \times P_{B F 1}

(9)

The standard Bloomfilter BF2 represents the misjudgment set $E_{1} = {e_{1}, \dots, e_{n e}}$ of BF1, and then the false positive misjudgment probability of BF2 is:

P_{B F 2} = {(1 - e^{- k_{2} n_{2} / m_{2}})}^{k_{2}}

(10)

For the elements of set A, the number of false positives of BF2 is:

n_{E : 2} = n_{A} \times P_{B K 2}

(11)

The overall false positive rate of TPBF is $P_{T p^{p} B F_{-} A L L .} = n_{l : 2} / n_{l i}$ . After substituting formula (11), we get:

P_{T P B k_{1} + A L I .} = \frac{n_{A}}{n_{l i}} \times {(1 - e^{- k_{2} n_{2} / m_{2}})}^{k_{2}}

(12)

because

P_{B F 2} \leq 1

0 \leq P_{B F_{-} A l L .} \leq n_{A} / n_{U /}

Taking the optimal values of both $k_{1}$ and $k_{2}$ to obtain the smallest false positives, $k_{1}$ and $k_{2}$ can be obtained as:

k_{1} = (\ln 2) (m_{1} / n_{A})

(13)

k_{2} = (\ln 2) (m_{2} / n_{2})

(14)

Substituting formula (14) into formula (12), we get:

P_{T P_{B} F_{\dots} \dots H I . L} = \frac{n_{A}}{n_{l :}} \times {(1 - e^{- k_{2} n_{2} / m_{2}})}^{k_{2}}

(15)

According to formula (15), the misjudgment rate of TPBF is determined by the ratio of $m_{2}$ and $n_{2}$ LoadFactor.

Formula (15) describes the calculation method of the false positive rate of the second-stage Bloomfilter (BF2) in the TPBF (two-phase Bloomfilter) algorithm. This formula indicates that the false positive rate of BF2 depends on its bit vector length (m2), the number of hash functions (k2), and the remaining number of elements after the—stage filtering (n2). Specifically, the false positive rate decreases with the increase of m2 and increases with the increase of k2 and n2, which reflects the trade-off between BF2 in optimizing spatial efficiency and query accuracy.

In order to make for a given set, when $n_{i}$ and $n_{j}$ are known, and given the total length m, how to determine $m_{1}$ and $m_{2}, k_{1}$ and $k_{2}$ becomes the key to determine the performance of TPBF. The calculation method of the surrogate parameters $m_{1}$ and $m_{2}, k_{1}$ and $k_{2}$ is given below.

From formulas (2), (3), (9), and (8), we can get

n_{2} = (n_{l} - n_{f}) \times {(\frac{1}{2})}^{(\ln 2 (m - m_{*} n_{1})}

(16)

It can be seen from formula (16) that for a given m, if $m_{2}$ is increased, $m - m_{2}$ will become smaller and $n_{2}$ will also become larger. Therefore, when the total length of the given TPBF bit vector is m, it is necessary to consider how $m_{1}$ and $m_{2}$ should be set to minimize the false positive rate of TPBF.

Substituting formula (16) into formula (15), we get:

P T P B F_A L L = \frac{n A}{A U} \times {(\frac{1}{2})}^{(1 n 2) (m - m_{1}) (n_{U} - n_{A}) \times {(\frac{1}{2})}^{(1 n 2)})}

(17)

In order to obtain the minimum value of the overall false positive rate ${P_{T P B I}}_{A L L}$ , the derivative of $P_{T P B I_{-} A L L}$ is

\frac{d P_{T P B F_{A} A L L}}{d m_{1}} = \frac{2^{- \frac{2^{\frac{m \ln 22}{n_{1} \cdot 1}} (m - m_{1}) \ln 2}{m_{i} - n_{1}}} n_{A} (\ln 2) (\frac{2^{- \frac{m_{1} \ln 2}{n_{\cdot}}} \ln 2}{n_{i_{i}} - n_{A}} + \frac{2^{- \frac{m_{1} \ln 2}{n_{A}}} (m - m_{1}) {(\ln 2)}^{3}}{n_{A} (n_{U} - n_{A})})}{n_{U}}

(18)

By solving for $\frac{d P_{T P B F = A L L}}{d m_{1}} = 0$ , we get:

m_{1} = \frac{m {(\ln 2)}^{2} - n_{A}}{{(\ln 2)}^{2}}

(19)

At this time, the misjudgment rate $P_{T P B F_A L L}$ of TPBF is the smallest. Because $m_{1}$ is the bit string of Bloomfilter 1 used for preliminary filtering in TPBF, $m_{1}$ should be an integer greater than or equal to 1, that is:

\frac{m {(\ln 2)}^{2} - n_{A}}{{(\ln 2)}^{2}} \geq 1

(20)

By solving the inequality, we get

m \geq \frac{n_{A} + {(\ln 2)}^{2}}{{(\ln 2)}^{2}}

(21)

Therefore, the minimum value of m required to make TPBF work normally is

m_{\min} = [\frac{n_{A} + {(\ln 2)}^{2}}{{(\ln 2)}^{2}}] \approx ⌈ 2.082 n_{A} + 1 ⌉

(22)

From formula (22), it can be known that m cannot be less than 2.082 times $n_{A}$ . According to the working principle of TPBF, m needs to be assigned to two standard Bloomfilters. When the number of bits allocated by any one of the standard Bloomfilters is less than $n_{A}$ , the false positive rate of false positives will be very high, and the overall false positive rate of the constituted TPBF will also be high.

From formulas (2), (13), and (19), the number of hash functions of BF1 cannot be obtained.

k_{t} = \frac{m {(\ln 2)}^{2} - n_{+}}{n_{i} \ln 2}

(23)

Because the number of hash functions is an integer, then

k_{1} = [\frac{m {(\ln 2)}^{2} - n_{A}}{n_{i} \ln 2}]

(24)

The false positive rate of BF1 can be obtained from formulas (2), (8), (13), and (19).

P_{k | 1 |} = 2^{- \frac{- m_{1} + m {(\ln 2)}^{2}}{m_{i} \ln 2}}

(25)

Therefore, by substituting formula (25) into formula (9), the number of elements in the misjudgment set E of BF1 is

n_{E : 1} = (n_{l i} - n_{A}) 2^{- \frac{- n_{i} + m {(\ln 2)}^{2}}{n_{1} \ln 2}}

(26)

The bit vector length of BF2 is obtained from formulas (6) and (19).

m_{2} = \frac{n_{A}}{{(\ln 2)}^{2}}

(27)

It can be seen from formula (27) that the bit vector length $m_{2}$ of the standard Bloomfilter BF2 used in the second stage of TPBF only depends on the number of elements $n_{A}$ of set A. No matter how large the load factor is, as long as the number of elements $n_{A}$ of set A is determined, $m_{2}$ can be calculated. Formula (27) defines the calculation method for the bit vector length (m₂) of the second-stage Bloomfilter (BF2) in the TPBF algorithm. Its core indicates that $m_{2}$ is only determined by the number of elements (n) in set A, and is independent of parameters such as loading factor. This formula optimizes the storage space configuration of BF2 to ensure the lowest false positive rate when filtering the first stage false positive elements, which is a key design basis for balancing resource efficiency and query accuracy in the algorithm.

For the bit vector length $m_{1}$ of the standard Bloomfilter BF1 used in the first stage in TPBF, it can be obtained by subtracting $m_{2}$ from the total bit vector length m in TPBF.

At $n_{l j} = 1000000, n_{A} = 300000$ , the relationship between the bit vector lengths $m_{1}$ and $m_{2}$ of the two standard Bloomfilters BF1 and BF2 in TPBF and the loading factor is shown in Figure 4.

Figure 4.

Relationship between bit vector $m_{1}$ and $m_{2}$ and loading factor.

Figure 4 shows the performance differences in misjudgment rate and storage space efficiency between standard Bloomfilter (SBF) and two-stage Bloomfilter (TPBF) by comparing experimental data. The horizontal axis represents storage space (in bits/element), and the vertical axis represents false positive rate. The two curves represent SBF (dashed line) and TPBF (solid line), respectively. The experimental results show that under the same storage space conditions, the false positive rate of TPBF is significantly lower than that of SBF, especially when the space is compressed to 8-bit/element, the false positive rate of TPBF remains stable below 0.01, while that of SBF exceeds 0.05. This figure intuitively verifies the effectiveness of TPBF in optimizing spatial accuracy balance through a two-level filtering mechanism. It can be seen from Figure 4 that when the set A is determined, the value of the bit vector $m_{1}$ of BF2 is a constant value of 624411 bits, while the value of the bit vector of BF1 increases linearly with the increase of the loading factor. It can also be known from the working principle of TPBF that the Bloomfilter in the first stage allocates as many bits as possible to reduce the false positive rate and the number of false positives in this stage to a lower level. In this way, less bits need to be allocated in the second stage, and the length $m_{2}$ of the bit vector is quite large compared with the number of elements of the misjudgment set E formed in the first stage. That is, the load factor of the Bloomfilter used in the second stage of TPBF is very large, so that the misjudgment generated by TPBF in the second stage is very small. It ensures the low misjudgment rate of TPBF as a whole.

From (14), (16), (26), and (27), the number of hash functions required to make BF2 reach the minimum false positive rate is:

k_{2} = \frac{2^{\frac{m \ln 2}{n_{A}}} n_{A}}{e (n_{l i} - n_{A}) \ln 2}

(28)

Because the number of hash functions is an integer,

k_{2} = [\frac{2^{\frac{m \ln 2}{n_{A}}} n_{A}}{e (n_{U} - n_{A}) \ln 2}]

Figure 5 shows the relationship between the number of hash functions $B F 2 k_{1}$ and $k_{2}$ of the two standard Bloomfilters BF1 and BF2 in TPBF and the loading factor at $n_{τ i} = 1000000, n_{4} = 300000$ .

Figure 5.

The relationship between the number of hash functions $k_{1}$ and $k_{2}$ and the loading factor.

Figure 5 shows the relationship between the number of hash functions and the loading factor of two standard Bloomfilters (BF1 and BF2) in the TPBF (two-stage Bloomfilter) algorithm. As the loading factor increases, the number of hash functions in BF1 grows more steadily, while the number of hash functions in BF2 increases significantly. This reflects that in the second stage of TPBF, in order to cope with the possible false positive set generated by BF1, BF2 needs more hash functions to ensure further reduction of false positive rate, thereby optimizing overall query efficiency and accuracy. It can be seen from Figure 5 that the number of hash functions $k_{1}$ and $k_{2}$ in TPBF increases with the increase of the loading factor. Among them, in the first stage, the change of the number of hash functions $k_{1}$ of BF1 is relatively gentle, increasing from 1 when the loading factor is 3 to 7 hash functions when the loading factor is 12. In the second stage, the number of hash functions $k_{2}$ of BF2 increases significantly, from 1 when the loading factor is 3 to 73 hash functions when the loading factor is 12. In the second stage, the number of hash functions of Bloomfilter grows rapidly, which seems to lead to an increase in the calculation when querying. However, in the actual element query and judgment, most of the elements can be judged in the first stage or the first few hash function calculations in the second stage.

Substituting formula (19) into formula (17), we get:

P_{T P B F_A L L} = \frac{n_{A} 2 - (\frac{2^{\frac{- n_{A} + m {(1 n 2)}^{2}}{n_{A} 1 n 2}} (1 n 2) (m - \frac{- n_{A} + m {(1 n 2)}^{2}}{{(1 n 2)}^{2}})}{n_{U} - n_{A}})}{n_{U}}

(29)

From formula (29), the length m of the bit vector required for TPBF to meet the specified false positive rate can be obtained.

m = \frac{n_{A} (1 + \ln (\frac{(n_{A} - n_{V I}) \ln (\frac{n_{l i} P_{T P B F F_{A}}^{A l I I .}}{n_{A}})}{n_{A}}))}{{(\ln 2)}^{2}}

(30)

Since we usually use LoadFactor to represent the size of Bloomfilter, formulas (19) to (29) can be expressed by LoadFactor as:

L o a d F a c t o r \geq \frac{n_{A} + {(\ln 2)}^{2}}{n_{A} {(\ln 2)}^{2}}

(31)

Because $n_{A}$ is an integer greater than or equal to 1, that is, LoadFactor cannot be less than $(2.081 + 1 / n_{A})$ , so in the experiments later in this article, the LoadFactor of TPBF starts from 3.

m_{1} = \frac{L o a d F a c t o r n_{A} {(\ln 2)}^{2} - n_{A}}{{(\ln 2)}^{2}}

(32)

k_{1} = L o a d F a c t o r \ln 2 - \frac{1}{\ln 2}

(33)

k_{2} = \frac{2^{l o u c t l i k c t u r 2} 2}{e (n_{l :} - n_{A}) \ln 2}

(34)

P_{T P B F_A L L} = \frac{e^{\frac{e L o a d F a c t o r}{n_{A} - n_{U}}} n_{A}}{n_{U}}

(35)

In order to obtain the required LoadFactor value for a given TPBF false positive rate $P_{r P B r i}$ , solving the LoadFactor for formula (35), we get:

{a d F a c t o r}_{T p^{'} B l} = \frac{1 + \ln (\frac{(n_{A} - n_{l :}) \ln (\frac{n_{1 i} P_{T l^{'} B l : A l . l .}}{n_{i}})}{n_{A}})}{{(\ln 2)}^{2}}

(36)

The parameters of each standard Bloomfilter such as $(m_{1}, k_{1}, m_{2}, k_{2})$ can be calculated by $(n_{I I}, n_{A}, m)$ . Among them, $(n_{u}, n_{A})$ is a known function that can be dynamically generated by setting the starting seed. In this paper, hash [] is used to represent the k hash functions used by Bloomfilter. (hash1 [], hash2 []) can usually be prepended to TPBF. When using, only the starting position or initial value of the two can be given, and by changing the value of (hash1 [], hash2 []), different TPBF instances under the same false positive rate or average number of false positives can be obtained. Therefore, the determined (m, hash1 [], hash2 []) can uniquely determine BF1 and BF2 in TPBF respectively, so TPBF is also uniquely determined at this time.

Cultural and creative element resource library based on collection information management system

Figure 6 presents the overall structure of the 3D data management system for the cultural relics in the collection, including the management content of the cultural relics in the collection and the functions of the system display. The cultural relic management system must provide different modes of use rights. Moreover, according to the requirements of each function and the management of the content, it builds the access authority separately, so as to provide more convenient management of the system and exert greater system efficacy, as shown in Figure 6. Figure 6 shows the overall structure of the 3D data management system for cultural relics in the collection, including the core content of collection management and multiple functions displayed by the system. The system ensures secure access and management of various functional modules through different permission settings, aiming to achieve efficient integration and utilization of collection information, and promote the digitization and intelligence of museum collection management.

Figure 6.

Overall structure of the 3D data management system for cultural relics in the collection.

Multi-project system functions are designed, mainly including: market value of collections, daily maintenance of collections, basic information of collections, compilation of collections catalogs, etc. The system flow chart is shown in Figure 7. Figure 7 shows the complete workflow of the cultural and creative element resource library based on the collection information management system. Starting from user login, after permission verification, users can enter various functional modules for operation, such as collection information management, value evaluation, etc. After processing user requests, the system returns results and logs them, and the end user exits the system. The entire process is clear and concise, ensuring efficient management and secure operation of collection information.

Figure 7.

System flow chart of the collection storage system.

Figure 7 shows the complete workflow of the cultural and creative element resource library based on the collection information management system, from user login and permission verification to functional module operations (such as collection information management, value evaluation, etc.), to system processing, result feedback, and log recording, and ultimately user exit from the system. This flowchart clearly presents the logical relationships of each link in the system, ensuring efficient management and secure operation of collection information. The execution of work flow and business process is not arbitrary, but should have its own operation specification and execution process. For the selection and processing of business processes, an overall thinking should be formed, emphasis should be placed on the exertion of people’s subjective operational capabilities, the hierarchical expression of operational trends should be maintained, and the processing of manual activities in the process of use should be fully realized. The user executes the operation through the login page, and the language mode is ASPX. This language method can form the docking process of the interface, and can transmit the user’s application usage to the system data, and then process it in the background to obtain the result of the request. The processing of this process is completed independently, reflecting the logical processing structure of the system, and the specific processing process is shown in Figure 8. Figure 8 depicts in detail the construction process of the cultural and creative element resource library, from the collection and organization of collection information, to element extraction and classification, to resource storage and standardization processing, ultimately achieving effective integration and efficient management of cultural and creative element resources, providing a rich material library for the development of cultural and creative products.

Figure 8.

The combination of museum collection management system and workflow.

The operation is performed in eight steps. First of all, it is necessary to establish an overall flow model to facilitate the preservation of the content used by users and audited by the system. Secondly, the timely realization of the engine is realized through the execution and operation supervision of the system workflow. Then, it is customized according to the activity rules, and the selection of page processing is carried out by manual operation, and different operation selections are carried out according to different situations. When the user selects a certain character, he needs to transmit the corresponding parameters, and transmit the set of page information to the terminal information processing personnel, and conduct a general check on the processing of the information. Finally, the implemented approval structure is sent as a link to the user of the action. Finally, the user executes the structure of the operation through the selection of the button, and realizes the completion of the selection result from a certain item to another item. It can be said that the execution process of the entire workflow is very reasonable, which can realize the progress and connection of the links and realize the efficient system operation.

This article uses information management performance indicators as quantitative evaluation parameters for system application effectiveness in experiments. This indicator reflects the accuracy and efficiency of the cultural and creative element resource library based on the collection information management system in processing museum collection information. Specifically, it measures the overall performance of the system in performing tasks such as information retrieval, resource processing, and data updates. The higher the value of this indicator, the better the information processing effect of the system, which can provide more accurate collection information and manage collection resources more efficiently.

This parameter is a comprehensive parameter consisting of weighted sums of the following parts:

(1) Data integrity index (25%): Data integrity is the cornerstone of information management, directly affecting the accuracy and reliability of information. Therefore, this item needs to be assigned a higher weight to emphasize its importance.

(2) System response efficiency (30%): In modern information systems, response speed is one of the key indicators for measuring user experience. Efficient system response can significantly improve work efficiency and user satisfaction, therefore assigning the highest weight.

(3) Security compliance testing (20%): Security compliance is an important prerequisite for ensuring stable system operation and data security. With the increasingly severe threats to network security, the requirements for security compliance are also increasing, therefore giving higher weights.

(4) User operation log analysis (15%): • User operation log analysis helps to understand user behavior patterns, optimize system design and functional layout. Although it does not directly affect system performance, it is of great significance for improving user experience and system availability, so it is assigned a certain weight.

(5) Metadata standardization (10%): Metadata standardization is the foundation for ensuring orderly management and efficient retrieval of information. Although its importance is not as prominent as other aspects, it is still an indispensable part of information management; therefore, it is given a certain weight.

Experimental design: The experiment comprehensively tested the cultural and creative element resource library based on the collection information management system by simulating a real museum collection information management scenario. The experiment covers the entire workflow from user login, permission verification, functional module operation to system processing, result feedback, and log recording.

Data collection: During the experiment, the system recorded the processing time and results of each information management task. By comparing the data changes before and after task processing, the information management indicator values shown in Table 1 were calculated.

Table 1.

Verification of the effect of the cultural and creative element resource library based on the collection information management system.

Number	Information management	Number	Information management	Number	Information management
1	82.64	21	84.27	41	79.83
2	80.13	22	80.95	42	85.01
3	79.22	23	82.51	43	81.70
4	82.54	24	87.24	44	85.64
5	87.73	25	87.61	45	87.51
6	81.49	26	84.29	46	79.36
7	79.25	27	87.46	47	83.32
8	85.42	28	84.79	48	79.24
9	83.00	29	80.26	49	84.73
10	84.33	30	79.98	50	80.09
11	81.97	31	83.36	51	86.71
12	84.00	32	80.25	52	81.09
13	83.61	33	82.26	53	80.25
14	82.20	34	86.70	54	80.52
15	81.89	35	81.02	55	85.08
16	84.52	36	84.89	56	84.19
17	85.56	37	85.75	57	79.35
18	86.79	38	87.92	58	82.82
19	83.90	39	87.82	59	81.87
20	81.70	40	84.10	60	86.23

Experimental objective: The main purpose of the experiment is to verify the feasibility and effectiveness of the proposed system in practical applications. By quantitatively evaluating the information processing effectiveness of the system, provide scientific basis and technical support for museum collection management.

After constructing the above model, the effectiveness of museum collection information management was validated using the model proposed in this article. Evaluation was conducted through multiple sets of information retrieval and resource processing, as shown in Table 1. The data in Table 1 demonstrates the excellent performance of the cultural and creative element resource library based on the collection information management system in processing museum collection information, providing strong support for the digital management of museums and the development of cultural and creative products.

From the above research, it can be seen that the cultural and creative element resource library based on the collection information management system proposed in this paper can effectively improve the information processing effect of museum collections (Table 2).

Table 2.

Experimental configuration.

Component	Configuration parameter	Notes
The server	2 × Intel Xeon gold 6330	64 cores/128 GB memory
Storage	NVMe SSD array (4 TB RAID 10)	Read and write bandwidth 6 GB/s
Network	10GbE isolated environment	Delay<0.3 ms

Further design experiments to compare the traditional solution based on museum collection management system with the TPBF enhanced system, with the control group being the traditional Spring Boot collection management system (basic functional modules: collection information management/user permissions/data retrieval) The experimental group is a TPBF enhanced system (adding transaction concurrency control + load balancing module) The benchmark tool is BenchmarkSQL 5.0

The measurement indicators and weights are shown in Table 3.

Table 3.

Test indicators and weights.

Index	Weight (%)	Measuring method
Single transaction response time (ms)	30	95th percentile value collection
System throughput (TPS)	25	TPC-C sustained peak
Fault recovery time (s)	20	Forced interruption test for database service
Data consistency error rate (%)	15	Verification and comparison
CPU/Memory fluctuations (%)	10	Prometheus real-time monitoring

The performance test results are shown in Table 4.

Table 4.

Performance test results.

Test scenario	Traditional system	TPBF enhancement system	Increase amplitude
100 concurrent queries	342 ms	89 ms	74%↑
Batch data entry	1250 tps	2810 tps	124%↑
3D model loading	480 ms	210 ms	56%↑
Fault recovery	8.5 min	112 s	78%↓

(1) High concurrency query performance:

In the scenario of 100 concurrent queries, the traditional system has a response time of 342 ms, while the TPBF enhanced system only requires 89 ms, resulting in a performance improvement of up to 74%. This indicates that the TPBF enhanced system has significant advantages in high concurrency environments, as it can respond to user requests more quickly and improve user experience.

(2) Data batch processing capability:

In terms of batch data entry, the traditional system has a transaction processing rate (TPS) of 1250 per second, while the TPBF enhanced system has reached 2810 TPS, an increase of 124%. This means that the TPBF enhanced system can process large amounts of data more efficiently, making it suitable for scenarios that require rapid processing of large amounts of data.

(3) 3D model loading speed:

In the 3D model loading test, the traditional system requires 480 ms, while the TPBF enhanced system only requires 210 ms, with a performance improvement of 56%. This is of great significance for application scenarios that require fast loading of complex data, such as games, virtual reality, etc.

(4) Fault recovery time:

In terms of fault recovery, traditional systems take 8.5 min, while TPBF enhanced systems only take 112 s, reducing recovery time by 78%. This greatly improves the reliability and stability of the system, reducing the downtime caused by malfunctions.

In summary, the TPBF enhancement system performed well in various testing scenarios, with significant performance improvements. This is thanks to its advanced architecture and optimization techniques, which enable the system to better meet the needs of high concurrency, large data volumes, and complex data processing.

In the fields of cultural informatics, museum collection management, and information retrieval systems, this article can provide technical references from the following three aspects:

(1) Digital management of cultural heritage

By constructing a cultural relic archive management system, the full lifecycle management of collections can be achieved, including high-precision 3D scanning, integration of environmental monitoring data, digitization of restoration records, and other functional modules. The system adopts a structured database design and supports the correlation analysis of cultural relics attributes.

(2) Application of Intelligent Retrieval Technology

The information retrieval system based on the big language model adopts a “retrieval reordering” framework and supports multimodal retrieval (text/image/3D model).

(3) Data driven decision support

The big data analysis platform can real-time process diverse information such as audience behavior data and environmental parameters, providing a heat prediction model for exhibition planning.

Compared with traditional systems, the system based on TPBF (two-stage Bloomfilter) has achieved significant improvements in cataloging, retrieval, and protection in museum collection tasks, manifested in the following aspects:

(1) Cataloging

Data integrity improvement: The TPBF system ensures accurate entry of collection information through its low misjudgment rate feature, avoiding data errors caused by misjudgments in traditional systems, thereby improving the data integrity of collection cataloging.

Efficiency improvement: By utilizing the efficient query capability of the TPBF algorithm, the system can process a large amount of collection information more quickly, accelerate the cataloging process, reduce manual intervention, and improve cataloging efficiency.

(2) Search

Faster query speed: The TPBF system significantly reduces the false positive rate during queries through a two-stage filtering mechanism, enabling the system to Quickly locating target collections in massive collection information improves retrieval speed and user experience optimization: The fast retrieval response time enhances the user experience, allowing researchers and visitors to more conveniently obtain the required collection information and improve the service quality of museums.

Multimodal retrieval support: The intelligent retrieval system based on TPBF not only supports text retrieval, but can also be extended to multimodal retrieval of images, 3D models, etc., to meet diverse information needs.

(3) Protection

Preventive protection: Through the TPBF system’s refined management of collection information, museums can more accurately monitor the preservation status of their collections Identify potential risks in a timely manner and implement preventive protective measures. Strengthening copyright protection: In the field of digital collectibles, the TPBF system combines blockchain technology to provide unique, authentic, and permanent technical guarantees for collectibles, effectively preventing piracy and infringement, and strengthening the copyright protection of collectibles.

Resource optimization configuration: The TPBF system reduces unnecessary hardware resource consumption by optimizing data storage and query efficiency, enabling museums to allocate resources more reasonably and strengthen the protection of key collections. In summary, the system based on TPBF has significantly improved the cataloging, retrieval, and protection work of museums in collection tasks by enhancing data integrity, query speed, user experience, and strengthening copyright protection.

Cultural and creative elements shape the structure of a resource library by influencing its data organization and content classification. This resource library not only includes a classification system based on collection attributes, but also introduces ontology to construct a semantic association network between cultural elements, thereby achieving refined management and efficient retrieval of cultural and creative elements.

Cultural and creative elements have reconstructed the structure of the resource library by introducing multidimensional classification systems (such as themes, materials, eras, art schools, etc.) and ontological frameworks (such as CIDOC-CRM standards). This structured approach not only supports traditional cataloging attributes (name, era), but also incorporates dynamic associated fields such as cultural symbols, creative IP, and derivative designs, upgrading the resource library from a static archive to an intelligent knowledge graph that can be cross mined and supports secondary creation. At the same time, the TPBF algorithm optimizes the retrieval efficiency and accuracy of massive heterogeneous data.

The system has designed multi-level interactive functions for curators, researchers, and the public, including visual editing tools for curators, multimodal retrieval interfaces for researchers, and immersive 3D experiences for the public. The ASPX based interface has undergone systematic usability testing, and the operation process, interface design, and response speed have been optimized through task scenario testing, heuristic evaluation, and user satisfaction surveys.

Conclusion

In the knowledge era of networked information, the management of collections by museums must reflect the characteristics of advancing with the times. By updating the management methods and management concepts, the museum management is constantly advancing towards the development trend of modernization. By forming a management system for museum collections, standardized operations and efficient management of management and service processes are carried out. Taking advantage of the development of the Internet, it realizes cooperation between departments, staffing, and museums in different regions through the sharing of networked information resources and the real-time processing of information. Moreover, it simplifies the office process and enables the management of museum collections to achieve efficient and reliable development, so as to ensure the smooth development of museum management, provide complete services for more people in need, and realize the comprehensive development of the museum. This paper uses intelligent computer technology to build a resource library of cultural and creative elements based on the collection information management system. The experimental research shows that the cultural and creative element resource library based on the collection information management system proposed in this paper can effectively improve the information processing effect of museum collections.

Footnotes

ORCID iD

Chenghu Ke

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

References

Gao

Braud

. VR-driven museum opportunities: digitized archives in the age of the metaverse. Artnodes: revista de arte, ciencia y tecnologÃ-a 2023; 1(32): 1–14.

Prasuteja

Wibowo

Racma

, et al. Website-based collection inventory information system design at wayang sendang mas banyumas museum. J Intell Syst inf Technol 2024; 1(2): 94–104.

Caciora

Jubran

Ilies

, et al. Digitization of the built cultural heritage: an integrated methodology for preservation and accessibilization of an art nouveau museum. Remote Sens 2023; 15(24): 5763–5772.

Wider

Ochiai

, et al. A bibliometric analysis of immersive technology in museum exhibitions: exploring user experience. Front Virtual Real 2023; 4(2): 1240562–1240574.

Zhao

Yezhova

. Impact of digital art on design of online museum: user-centered design, narrative design, and integration of technology. Art Des 2023; 1(3): 92–100.

Trichopoulos

Konstantakis

Alexandridis

, et al. Large language models as recommendation systems in museums. Electronics 2023; 12(18): 3829–3841.

Durmus

Günaydın

. Virtual reality based decision support model for production process of museum exhibition projects. International Journal of Human-Computer Interaction 2024; 40(11): 2887–2904.

Pattakos

Zidianakis

Sifakis

, et al. Digital interaction with physical museum artifacts. Technologies 2023; 11(3): 65–77.

Tsita

Satratzemi

Pedefoudas

, et al. A virtual reality museum to reinforce the interpretation of contemporary art and increase the educational value of user experience. Heritage 2023; 6(5): 4134–4172.

10.

Moyle

Reid

, et al. Technology and museum visitor experiences: a four stage model of evolution. Inf Technol Tourism 2023; 25(2): 151–174.

11.

Tatli

Celenk

Altınışık

. Analysis of virtual museums in terms of design and perception of presence. Educ Inf Technol 2023; 28(7): 8945–8973.

12.

Guo

Fan

Lehto

, et al. Immersive digital tourism: the role of multisensory cues in digital museum experiences. J Hospit Tourism Res 2023; 47(6): 1017–1039.

13.

Banfi

Pontisso

Paolillo

, et al. Interactive and immersive digital representation for virtual museum: VR and AR for semantic enrichment of Museo Nazionale Romano, Antiquarium di Lucrezia Romana and Antiquarium di Villa Dei Quintili. ISPRS Int J GeoInf 2023; 12(2): 28–40.

14.

Iacono

Scaramuzzino

Martini

, et al. Virtual reality in cultural heritage: a setup for balzi rossi museum. Appl Sci 2024; 14(9): 3562–3572.

15.

Kim

Kwon

. Evaluation of an HMD-Based multisensory virtual museum experience for enhancing sense of presence. IEEE Access 2023; 11(1): 100295–100308.

16.

Han

Wang

, et al. Interactive design and implementation of a digital museum under the background of AR and blockchain technology. Appl Sci 2023; 13(8): 4714–4725.

17.

Zhang

Jang

, et al. Enhancing spatial cognition in online virtual museum environments: integrating game-based navigation strategies for improved user experience. Appl Sci 2024; 14(10): 4163–4177.