Intelligent quranic story builder

1. Introduction

The lives of Holy prophets are role model, perpetual guidance, and complete code of conduct for entire humanity from the Adam till the last messenger, the prophet Mohammad (PBUH). Even the modern secular theory is borrowed from the holy wisdom. The collection of information about their lives, personal conduct, teachings, social uplifting, law and order rules were the most sacred and precious duty for the followers of all religions, and they truly amazed the humanity to compile the information meticulously. Therefore, this subject matter remained, and still is considered, not less than a blessing even in era of secular theories, science and information. Astonishingly, the modern tools and technologies are the most favorite and promising tools for preaching and propagation of such sublime thoughts. But, an enthusiastic reader is often hampered when the information and teaching of all, particularly the main divine religion is found scattered in multiple forms and destinations.

For example, in holy Quran, the story of Musa starts in 2 nd surah (Al-Baqra), wishing blessings upon the Bani Israel and, is scattered in the entire Quran, and occurs as example and comparison among the believers of both religions. Christians believe in Jesus and Holy Book New Testament (Injeel Muqadas). Whereas, in New Testament, the lives of all Holy Prophets from Adam to Jesus bewilder in the entire Holy Bible. Similarly, Jewish believe in the Muses, and Holy Book Old Testament (Ahdnama Ateeq). According to Old Testament the lives of Holy Prophets start from Adam, Eve and his sons Sheet and Able, and this all description is not a sequence.

The modern man wants it all be organized at one plate form, but unfortunately, hardly there exists any destination in soft or hard fields containing such information and organization. It is still a mega challenge for the technology and experts to evolve a system to stem all in one section and a shared platform. To negotiate with such quizzical task, it was required drastically to have such a system on computer which may trace, collect, pattern, arrange and finally present the information about the pious personalities in form of logical stories. Since obsessed by this scholarly and technological impulsion this research has been proposed and successfully implemented. This application by applying Fuzzy Cognitive Map (FCM), builds a Semantic Linked Network (SLN), which collects information about the lives of prophets, regions, teaching, conflicts, family, state and impact on the discourse of history from the Holy Corpus, and states in a form of logical narrative. People can observe the story of lives of Holy Prophets automatically based on keywords, themes, frequency of keywords and weightage.

Our proposed system works in two steps. At first it collects all related information from different resources by building a semantic network through fuzzy logic for a concept (keyword for which individual needs a story like Muses). In the second step, a story is generated by some expert based on collected information. Proposed system is semi-automatic because first step is automatic and second step involves religious expert. Prototype for the proposed system is implemented and presented in this research. In this development, Fuzzy Cognitive Map (FCM) data mining techniques for extracting the data is implemented. The main source of this research is the Holy Corpus, that is inactive dataset and in plain-text form. Following steps are carried out to build the proposed system.

•

Enlisting the concept (keywords)

•

Finding the frequency on element basis

•

Finding the frequency on section basis

•

Finding the frequency on document basis

•

Finding the concept value of each concept using the fuzzy formula

•

Finding the co-occurrence/s

•

Finding the weight of each link using Fuzzy formula

•

Building SLN from Holy Corpus

•

Compilation of results in the form of story

The remaining sections are devoted in such a way; related literature in section-II. Section-III, covers the description of the proposed system for semantic story building. Whereas, Section-IV, has been allocated to results and discussions. Lastly, section-V, pertains to conclusions.

2. Related work

Data Mining is the major process of discovering knowledge. It focusing upon different Methodology for extracting useful knowledge from data and there are several useful knowledge and data discovery (KDD) tools to extract the knowledge. Following is a review of such systems and techniques, well-known in the literature.

In [1], authors designed a smart E-service system applying the fuzzy technique to build to evaluate the perfection of quality of services FCM [1], the built system appeared as one of the excellent contribution in arena of electronic. It possessed three distinguished features, ‘strong’, ‘weak’, ‘moderate’ through fuzzy weight. The system was designed to sustain the relationship between producer and consumer to enrich quality and service. Luo et al. [2] engineered a semantic representation system to excel version of FCM using matrix system. Utilizing the keywords, co-occurrence and weightage of same word for building the SLN for information relating the medical science with the help of keywords weight, represented basic method of FCM and provide solution in the form of Scientific Representation of Documental Fragmentations (SRDF) and then drop the noising words like (the, of, is etc.) and built strong semantic link in medical field data. In [3], authors represented the flow of web for the users help. How a user can get the relevant and correct information, this is the basic problem. To solve this problem an intelligent system was built for creating strong relations. However, this system is domain specific not general.

In [4], the authors worked on intelligent browsing for specific topics and discovered related topics of the keywords. They found that intelligent browsing was a key issue in the field of knowledge grid during exploring on web as it shows a lot of irrelevant data. For that purpose, they used an enhanced technique of FCM for building a powerful association with keywords. Authors of [5] worked on An Incremental Semantic Data Model for Organization of Web Resources in China. They used association link network for that purpose by aiming at extended hyperlink network to an association link network. A Naïve way was applied to find out the time complexity for building the network. Zhuge and Luo worked on automatic generation of relative keywords in the field of Electronic [6]. Authors used simple FCM with adjacency matrix by passing through all the three steps and various necessary steps to produce strong entity for semantic links. In this purpose, they used keywords, frequency, co-occurrence and weightage using Element Fuzzy Cognitive Map (E-FCM), Section Fuzzy Cognitive Map (S-FCM), and Documents Fuzzy Cognitive Map (D-FCM). The experiment was done in web domain.

In [8], the authors experimented on enhanced version of Apriori algorithm for solid and strong association rule mining. They selected sales transaction domain for experiment and implemented a new algorithm namely AprioriHybrid. That Algorithm had developed features in relation with different transaction sizes and various numbers of items on a big database. In [9], authors worked on data searching and exploring the knowledge on web in a new way by using Association Link Network (ALN) instead of FCM for building of Semantic Link, this system provide three steps for building of ALN: C-ALN for basic level as connection rich; K-ALN for medium level as Kernel level and O-ALN as Original ALN. In C-ALN data is refined from irrelevant data at basic level. In K-ALN data became more accurate while O-ALN was a desired semantic and built the closest relation of the web resource. In this approach system work at basic level to achieving the optimum level and gradually dropped the loose relation between the words. Hasegawa et al. [9] found the important relations in a large document. This discovery help out users to not only for understanding the text but also for making the summery of large scale data.

Zhu et al. [10] worked on ALN [11] to provide a new algorithm for making a relation on web resources on large scale active resources management and make a rich and optimum link with the help of new formula This formula was applied to rapidly and accurately acquire incremental data in previous existing web resources. Xu et al. [11] worked on organization of Large Scale Online news reports via system based on semantic link network.

Shahzadi et al. [13, 14, 15] engineered graphical classifier for Holy Quran using Semantic Network of Holy Quran and Holy Books. The knowledge is not organized in sequential way it is rather random or event based. The reason for building this intelligent tool was to help religious scholars who may be searching for concept throughout the Holy Quran. This research set up the first step for us to build the current system.

In [16], authors proposed a teacher assessment and profiling system (TAPS) for classifying the teachers according to their feedback and various other factors extracted from the learning management system (LMS). In [17, 18], authors presented a mechanism for user behavior classification of a network user based on his/her activities over the time. In this regard, a fuzzy rule based system was used as the classifier and data mining approach was used to find the trends of users, temporally.

In [19, 20], authors presented a study of students’ performance analysis and selection of appropriate teachers for specific subject and specific class based on his/her feedback, students’ grades, age group and various other attributes. Famous, Apriori algorithm and frequent pattern tree (FP-tree) algorithm were investigated for sake finding the hidden truth in the raw data after converting it to a specific format by utilizing well known data fusion and cleaning techniques.

3. The proposed approach

For building of Intelligent Quranic Story Builder (IQSB), firstly all the three Holy Books i.e., Holy Bible, Holy Quran and Hadith were required in plain text form. All the keywords present in this Holy Corpus were gathered manually for creation of their stories, and then built a semantic linked network through FCM. First, a user creates or makes a list of keywords for required story, then save it on a database and search any word from a keyword list. IQSB built a SLN automatically and helps user to create a story with his/her preferences. Now the story is available in plain text form, which is saved on another database automatically, is ready for readers as one of the possible outcome of related searches.

Plain text document analysis, parsing and important concept extraction was the first step that was carried out. Unique words and their relationships were extracted from the text while removing the noise words and text segmentation.

The detailed process is depicted in Fig. 1.

OPEN IN VIEWER

Word or reserve word basically belong to a main story, that a user wants to build. First, a user finds the words/reserve words from relevant holy book. Or combine holy books namely holy corpus (Holy Quran, Holy Bible, Holy Hadith). Words/ reserve words are represented by K = { k 1 , k 2 , … , k n } . This set of reserve words is further divided into five subsets given as follows:

i. K Q = { k Q ⁢ 1 , k Q ⁢ 2 , … , k Q ⁢ q }

ii. K B = { k B ⁢ 1 , k B ⁢ 2 , … , k B ⁢ b }

iii. K H = { k H ⁢ 1 , k H ⁢ 2 , … , k H ⁢ h }

iv. K C = { k C ⁢ 1 , k C ⁢ 2 , … , k C ⁢ c }

v. K D = { k D ⁢ 1 , k D ⁢ 2 , … , k D ⁢ d }

Where K Q , K B , K H is the set of reserve words is taken only from Holy Quran, Holy Bible, Holy Hadith respectively. K C is the set of common reserve words, those are taken from Holy Corpus. K D is the set of different reserve words, those are not common in Holy Corpus.

After defining/extracting reserve words, find the frequency of those reserve words. Number of occurrence of a particular reserve words in a given database is called the frequency of that word, then find the state value of reserve word automatically.

Every reserve word has its own state value], S = { s 1 , s 2 , … , s n } the meaning of state value is own worth of reserve word. After finding this co-occurrence is required as completion of FCM [1, 2] C = { C 1 , C 2 , … , C n } where C i = { C i ⁢ 1 , C i ⁢ 2 , … , C i ⁢ n } and in combine fashion, C can be represented as:

(1)

weightage of that co-occurrence of reserve word is required W = { W 1 , W 2 , … , W n } , where W j = { w j ⁢ 1 , w j ⁢ 2 , … , w j ⁢ n } , and in matrix form can be seen as:

(2)

Now, all material is ready for calculation, according to FCM formula [1, 2] for automatic building of SLN with thrash-holding function:

Calculate ⁢ V c , j ⁢ ( t + 1 ) = threshold_func (3) ( ∑ i = n i = 1 i ≠ j ⁢ V c ⁢ i ⁢ ( t ) ⁢ w i ⁢ j )

Patches are available for building of strong story according to user wish, now user will arrange the patches for building the story using IQSB.

For experimental results, Android studio, Visual studio and XML graphs are used here to evaluate the proposed scheme. There are countless keywords in Holy Corpus, which is not possible to access in a single research manually. So, here n = 12 keywords are used for experiments.

In current research only three Holy prophets were selected including Prophet Adam (A.S.), Prophet Musa (A.S.) and Prophet Yousaf (A.S.) Prophet Adam (A.S.) and Prophet Musa (A.S.) both have 235 keywords related with their names while Prophet Yousaf (A.S.) has 373 keywords related to his name. Among all these keywords every word doesn’t have story (Creation, Tell, Rotten, Stale, Mud, Red, Sea, drowning, glad, fragrance, settled, embraced etc.). Around 120 stories were built in this research on different keywords (Table 1). Frequency and state values of all the keyword taken from Holy Bible were found out but sample of twelve keywords were presented in research work, the keyword Adam has (33, 0.9999999), God has (3883, 1), First has (13, 0.999655595), Tell has (27, 0.9999997), Satan has (55, 1), Heaven has (5, 0.96402758) respectively. Similarly, more words can be shown.

Their respective frequency and state values are also shown below.

Although, co-occurrence of almost all the keywords were found out in the software prototype but sample of only twelve keywords is presented here in this article work. The co-occurrence means the number of times a word appeared with the keyword. That is degree of togetherness, which shows a type of relevance between the pair of words.

These co-occurrences are given for keyword “Adam” in such a way that the keyword “Adam” with “Adam” has 0, “Adam” with “God” has 8, “Adam” with “First” has 1, “Adam” with “Tell” has 0, “Adam” with “Satan” has 3, “Adam” with “Heaven” has 0, “Adam” with “Special” has 0, “Adam” with “Gathering” has 0, “Adam” with “Together” has 0, “Adam” with “Creation” has 0, “Adam” with “first man” has 1 and “Adam” with “Eve” has 3 (Table 2).

Weightages of the respective experiment can be shown as given below in the table. The keyword Adam with Adam has 0.0000, Adam with God has 0.0021, Adam with First has 0.0769, Adam with Tell has 0.0000, Adam with Satan has 0.0000, Adam with Heaven has 0.0000, Adam with Special has 0.0000, Adam with Gathering has 0.0000, Adam with Together has 0.0000, Adam with Creation has 0.0000, Adam with first man has 0.3333 and Adam with Eve 0.0196 (Table 3).

OPEN IN VIEWER

In the story derived and developed from Holy Bible, the word “Eve” makes connection with “Adam”, whereas, contrary to this pattern, in Holy Quran, word “Hawa” relates with Adam. Similarly, when common keywords are selected from Holy Quran, Holy Bible and Holy Hadith “Eve” and “Hawa” appear as distinct word. Due to which such words cannot be assume common, yet the relation of “Hawa” and “Adam” sustained, by “together”. It proves that in this semantic linked network the names however are not common, does not affect the relationship. This experiment is evident that our semantic linked network has been constituted on that strong roots/foundations.

Now same experiment is repeated for Quran and results are discussed. Keywords can be shown below in Table 4. Frequency and state values of all the keyword taken from Holy Quran were find out but Again sample of twelve keywords were presented in research work, the keyword Adam has (25, 1) Satan has (32, 1), Cry has (10, 0.997458), together has (56, 1), creation has (52, 1), hawa has (24, 1), earth has (399,1), angles has (85,1) man has (468,1), fall (59,1) respectively etc.

Now the co-occurrence of Quranic keywords shown below and can be calculated as discussed above in Table 5.

Respective weightages data also shown below in weightage table (Table 6).

Node to node SLN building shown below. Relationship between “Adam”, “fall”, “together”, “earth” and “Hawa” shows the story that Adam and Hawa together fall down to earth. This is shown in Fig. 3.

OPEN IN VIEWER

The SLN comprising on 235 keywords approximately has been shown above from which eleven keywords have been sorted experimentally, which make a small SLN as shown in Fig. 4, from this mega SLN.

OPEN IN VIEWER

Intelligent Quranic Story Builder (IQSB) is an innovative approach which is combination of domain (Holy Corpus) and Fuzzy Cognitive Map (FCM). FCM in combination with Semantic Link Network (SLN) helps users to build adaptive Quranic stories. Each phrase of story can be authenticated by the relevant references. The stories are not present in an integrated form in Holy corpus. IQSB helps the user to read the stories spread in patches in entire Holy corpus in a single platform with an integrated view. Furthermore, building of SLN technique, such as FCM, ALN etc. was not applied before in this religious domain. So, in present study, both techniques and domain are used together. In IQSB, user can put either sort of theme concept related to these three holy books and find out the node to node SLN, in which every node possesses its unique story built in IQSB.